KO KNOWLEDGE ORGANIZATION Contents

Knowl. Org. 47(2020)No.3

Official Journal of the International Society for Knowledge Organization ISSN 0943 – 7444 International Journal devoted to Concept Theory, Classification, Indexing and Knowledge Representation

Contents

Articles Reviews of Concepts in Knowledge Organization

Luís Miguel Oliveira Machado, Maurício Barcellos Rick Szostak. Almeida and Renato Rocha Souza. The Basic Concepts Classification (BCC) ...... 231 What Researchers are Currently Saying about Ontologies: A Review of Recent Web of Science Björn Hammarfelt. Articles ...... 199 Discipline ...... 244

Karim Tharani. Arthur Smith. Just KOS! Enriching Digital Collections with Physics Subject Headings (PhySH) ...... 257 Hypertexts to Enhance Accessibility of Non-Western Knowledge Materials in Libraries ...... 220 Torjus Midtgarden. Peirce’s Classification of the Sciences ...... 267

Books Recently Published ...... 279 Knowl. Org. 47(2020)No.3

KNOWLEDGE ORGANIZATION KO

Official Journal of the International Society for Knowledge Organization ISSN 0943 – 7444 International Journal devoted to Concept Theory, Classification, Indexing and Knowledge Representation

KNOWLEDGE ORGANIZATION José Augusto Chaves GUIMARÃES, Departamento de Ciência da Informacão, Universidade Estadual Paulista–UNESP, Av. Hygino Muzzi This journal is the organ of the INTERNATIONAL SOCIETY FOR Filho 737, 17525-900 Marília SP Brazil. E-mail: [email protected] KNOWLEDGE ORGANIZATION (General Secretariat: Athena SALABA, School of Information, Kent State University, PO Box 5190, Michael KLEINEBERG, Humboldt-Universität zu Berlin, Unter den Kent, Ohio, 44242-0001, USA. E-mail: [email protected]. Linden 6, D-10099 Berlin. E-mail: [email protected]

Editors Kathryn LA BARRE, School of Information Sciences, University of Illinois at Urbana-Champaign, 501 E. Daniel Street, MC-493, Champaign, IL Richard P. SMIRAGLIA (Editor-in-Chief), Institute for Knowledge 61820-6211 USA. E-mail: [email protected] Organization and Structure, Lake Oswego OR USA 97035. E-mail: [email protected] Devika P. MADALLI, Documentation Research and Training Centre (DRTC) Indian Statistical Institute (ISI), Bangalore 560 059, India. Joshua HENRY, Institute for Knowledge Organization and Structure, E-mail: [email protected] Lake Oswego OR USA 97035. Daniel MARTÍNEZ-ÁVILA, Departamento de Ciência da Informação, Peter TURNER, Institute for Knowledge Organization and Culture, Lake Universidade Estadual Paulista–UNESP, Av. Hygino Muzzi Filho 737, Oswego OR USA 97035. 17525-900 Marília SP Brazil. E-mail: [email protected]

J. Bradford YOUNG (Bibliographic Consultant), Institute for Knowledge Widad MUSTAFA el HADI, Université Charles de Gaulle Lille 3, URF Organization and Structure, Lake Oswego OR USA 97035. IDIST, Domaine du Pont de Bois, Villeneuve d’Ascq 59653, France. E-mail: [email protected] Editor Emerita H. Peter OHLY, Prinzenstr. 179, D-53175 Bonn, Germany. Hope A. OLSON, School of Information Studies, University of E-mail: [email protected] Wisconsin-Milwaukee, Milwaukee, Northwest Quad Building B, 2025 E Newport St., Milwaukee, WI 53211 USA. E-mail: [email protected] M. Cristina PATTUELLI, School of Information, Pratt Institute, 144 W. 14th Street, New York, New York 10011, USA. Series Editors E-mail: [email protected]

Birger HJØRLAND (Reviews of Concepts in Knowledge Organization), K. S. RAGHAVAN, Member-Secretary, Sarada Ranganathan Endowment Department of Information Studies, University of Copenhagen. E-Mail: for Library Science, PES Institute of Technology, 100 Feet Ring Road, BSK [email protected] 3rd Stage, Bangalore 560085, India. E-mail: [email protected].

María J. LÓPEZ-HUERTAS (Research Trajectories in Knowledge Heather Moulaison SANDY, The iSchool at the University of Missouri, Organization), Universidad de Granada, Facultad de Biblioteconomía y 303 Townsend Hall, Columbia, MO 65211, USA. Documentación, Campus Universitario de Cartuja, Biblioteca del Colegio E-mail: [email protected] Máximo de Cartuja, 18071 Granada, Spain. E-mail: [email protected] M. P. SATIJA, Guru Nanak Dev University, School of Library and Editorial Board Information Science, Amritsar-143 005, India. E-mail: [email protected] Thomas DOUSA, The University of Chicago Libraries, 1100 E 57th St, Chicago, IL 60637 USA. E-mail: [email protected] Aida SLAVIC, UDC Consortium, PO Box 90407, 2509 LK The Hague, The Netherlands. E-mail: [email protected] Melodie J. FOX, Institute for Knowledge Organization and Structure, Lake Oswego OR USA 97035. E-mail: [email protected]. Renato R. SOUZA, Applied Mathematics School, Getulio Vargas Foundation, Praia de Botafogo, 190, 3o andar, Rio de Janeiro, RJ, 22250- Jonathan FURNER, Graduate School of Education & Information 900, Brazil. E-mail: [email protected] Studies, University of California, Los Angeles, 300 Young Dr. N, Mailbox 951520, Los Angeles, CA 90095-1520, USA. Rick SZOSTAK, University of Alberta, Department of Economics, 4 E-mail: [email protected] Edmonton, Alberta, Canada, T6G 2H4. E-mail: [email protected]

Claudio GNOLI, University of Pavia, Science and Technology Library, via Joseph T. TENNIS, The Information School of the University of Ferrata 1, I-27100 Pavia, Italy. E-mail: [email protected] Washington, Box 352840, Mary Gates Hall Ste 370, Seattle WA 98195- 2840 USA. E-mail: [email protected] Ann M. GRAF, School of Library and Information Science, Simmons University, 300 The Fenway, Boston, MA 02115 USA. Yejun WU, School of Library and Information Science, Louisiana State E-mail: [email protected] University, 267 Coates Hall, Baton Rouge, LA 70803 USA. E-mail: [email protected] Jane GREENBERG, College of Computing & Informatics, Drexel University, 3141 Chestnut Street, Philadelphia, PA 19104 USA, E-mail: Maja ŽUMER, Faculty of Arts, University of Ljubljana, Askerceva 2, [email protected] Ljubljana 1000 Slovenia. E-mail: [email protected]

Knowl. Org. 47(2020)No.3 199 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

What Researchers are Currently Saying about Ontologies: A Review of Recent Web of Science Articles†

Luís Miguel Oliveira Machado*, Maurício Barcellos Almeida**, Renato Rocha Souza***

*Universidade de Coimbra, Departamento de Filosofia, Comunicação e Informação, Largo da Porta Férrea, 3004-530 Coimbra, Portugal, **Universidade Federal de Minas Gerais, Departamento de Teoria e Gestão da Informação, Av. Antônio Carlos, 6627 Campus Pampulha, Belo Horizonte, M.G., Brazil 30.330-240, ***Escola de Matemática Aplicada, Fundação Getúlio Vargas, Av. Praia de Botafogo, 190, 3º Andar, Botafogo, 22250-900 - Rio de Janeiro, RJ – Brasil,

Luís Miguel Machado is an information science PhD student at the University of Coimbra and an associated researcher at the Centre of 20th Century Interdisciplinary Studies. He holds a master’s degree in information science and a teaching degree in visual and technological education. His current research topic is the ontological approach in knowledge organization.

Mauricio B. Almeida, PhD, is currently Associate Professor of the Graduate Program in Knowledge Organization & Management at the School of Information Science, Federal University of Minas Gerais, Brazil. He is a visiting fellow at the State University of New York at Buffalo, US, and still collaborates with the Ontology Research Group working with ontologies, knowledge representation, knowledge organization and information systems.

Renato Rocha Souza is Visiting Researcher at the Austrian Academy of Science, Professor and Researcher at Funda- ção Getúlio Vargas (FGV), Brazil, and Collaborator Professor at the Universidade Federal de Minas Gerais (UFMG), Brazil. He is a visiting fellow at the University of Glamorgan, UK and is a senior visiting scholar at Co- lumbia University. His main research topics include knowledge organization systems, natural language processing and machine learning.

Machado, Luís Miguel, Mauricio B. Almeida and Renato Rocha Souza. 2020. “What Researchers are Currently Saying about Ontologies: A Review of Recent Web of Science Articles.” Knowledge Organization 47(3): 199-219. 62 references. DOI:10.5771/0943-7444-2020-3-199.

Abstract: Traditionally connected to philosophy, the term ontology is increasingly related to information systems areas. Some researchers consider the approaches of the two disciplinary contexts to be completely different. Others consider that, although different, they should talk to each other, as both seek to answer similar questions. With the extensive literature on this topic, we intend to contribute to the understanding of the use of the term ontology in current research and which references support this use. An exploratory study was developed with a mixed methodology and a sample collected from the Web of Science of articles published in 2018. The results show the current prevalence of computer science in studies related to ontology and also of Gruber's view suggesting ontology as kind of conceptualization, a dominant view in that field. Some researchers, particularly in the field of biomedicine, do not adhere to this dominant view but to another one that seems closer to ontological study in the philosophical context. The term ontology, in the context of information systems, appears to be consolidating with a meaning different from the original, presenting traces of the process of “metaphorization” in the transfer of the term between the two fields of study.

Received: 22 January 2020; Revised: 5 March 2020, 31 March 2020; Accepted: 8 April 2020

Keywords: ontology, definitions, works, articles

200 Knowl. Org. 47(2020)No.3 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

† We are very grateful to the two anonymous reviewers for their helpful comments. We also gratefully acknowledge the excellent editorial work of Dr. Richard Smiraglia.

1.0 Introduction can be seen as types of restricted ontologies. Several issues in classification, since ancient times, originated in metaphysi- The ontological approach to describe reality—or put an- cal problems. Discussions and answers to the metaphysical other way, the way we determine what is similar and what is problem of universals gave rise to several theories or lines of different—is reflected in the organizational systems that we thought that have influenced the way we classify until today. develop. Traditionally connected to philosophy, the term The idea of concepts came from Kant and others as a sort of ontology has been increasingly related to information sys- questioning regarding old Aristotelian theories. The role of tems areas. In these two academic communities, there are LIS in applying and improving methods and theories of researchers who consider the approaches as completely dif- classification is widely known. However, in order to in- ferent; while in the former there is speculation about the struct computers to make inferences and classifications, for- structures of the world, the latter is focused on concrete mal ontologies are needed. problems of modeling domains of knowledge in computa- It is not the purpose of this study to comprehensively dis- tional artifacts (Poli and Obrst 2010). This understanding sect the meaning of the term ontology, a subject that has will not be oblivious to the use of the term ontology to des- been widely addressed in several studies. Good examples of ignate a concrete artifact of computational engineering works that clarify the meaning of the term ontology and (Guizzardi 2007). Disagreeing with the separation between make the connection between its use in the two disciplinary the two communities, Poli (2010a) considers that they contexts, the philosophy area and those related to infor- should talk to each other, as both seek to answer similar mation systems, are Almeida (2013), Poli and Obrst (2010) questions. and Smith (2003). Other similar works but with a perspec- Within the field of knowledge organization (KO), the tive focused on the context of information systems are Al- analysis of studies on ontology has generated myriad opin- meida and Bax (2003), Gruber (2009), Guarino and Gi- ions. On the one hand, Soergel (1999) suggests that studies aretta (1995) and Guarino et al. (2009). Examples of works on ontology would be some sort of bogus enterprise, as they where the emphasis is on the use of the term in the philo- concern classification, and classification is something that sophical sense are Hennig (2008) and Poli (2010b). Another cannot be reinvented; for Gilchrist (2003), the use of the term recurring approach is the search for a clarification of the ontology in other fields to name a type of classification struc- uses of the term ontology in the KOS spectrum, e.g., Grun- ture is only an etymological issue. Currás (2004) believes the inger et al. (2008), Khazraee and Lin (2011), Kless et al. term ontology would result from well-known concepts from (2011) and Souza et al. (2012). Other studies present ontol- LIS applied to new technologies. In the same line as Soergel, ogy as a process for knowledge organization, such as Poli Dahlberg (2014) states that it is a sort of reinvention of the (1996) and Smiraglia (2014). There are also works in which wheel to take seriously new technological aspects of existing the term ontology arises associated with a new scientific area organizational processes, which introduces unfortunate des- with designations such as “applied ontology” (Smith 2013) ignations—such as ontologies—to what is well-known as a or “formal ontology” (Herre 2015). Finally, we highlight the knowledge organization system (KOS). On the other hand, comprehensive bibliometric study of ontology research that there were visions that identify the similarity between ontol- covers the period from 1900 to 2012. In this study, Zhu et ogy research and bibliographic classification, fostering coop- al. (2015, 47) stress the importance of Thomas Gruber “for eration, as for example Vickery (1997). Emphasizing the im- the establishment of the theory of ontology in scientific portance of the procedural aspect of ontology, as a study of fields” and his “commonly accepted” definition of ontol- what exists, Smiraglia (2014) considers this process one of the ogy. pillars in the development of any KOS. With the extensive and relevant literature on the subject, All the discussion since the 1990s demonstrates how on- of which we present only a small part, we consider a perti- tologies have become a core subject in the knowledge organ- nent issue to question the impact on current ontology re- ization field, as one can see in the themes and subthemes of search. Thus, seeking to contribute to the understanding of the ISKO international conferences over time. This interest this subject, we intend to verify on which authors and their seems to be natural, since people understand that ontologies respective works current researchers base their definitions as representational artifacts are grounded in classification of the term “ontology.” Specifically, we aim to: principles, which are a seminal theme in LIS theories. Hjør- land (2019) corroborates with this interest in explaining that there are a variety of classification systems in KO that Knowl. Org. 47(2020)No.3 201 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

i) collect definitions of ontology presented in the most forms presented and the semantic relationships between the recent peer-reviewed articles published in 2018 from unit’s components (Krippendorff 2004). In articles with a sample taken from the Web of Science (WoS); more than one definition, we used contextual elements to ii) identify the most cited authors and papers in these select which would represent the authors’ opinion. In cases definitions; and, where these elements were not sufficient to clarify the iii) analyze the variations of meaning of those defini- meaning, we resorted to the typology of the definitions, tions. considering as the most representative those without quotation, then indirect quotations and, finally, direct quota- In the next section, we describe the methodology used in the tions. study to then summarize its results and complement the Regarding the distribution by epistemological area, the material collected with appendices presented at the end. In five major categories of WoS were considered: “arts & hu- the fourth section, the central question of the study is ex- manities,” “life sciences & biomedicine,” “physical sci- plored in the discussion of the results. Finally, in the last sec- ences,” “social sciences” and “technology.” These categories tion, we come forward with potential repercussions of the group the 153 research areas with which WoS articles are detected trend in current research. classified. In the case of articles included in more than one category, the category with the most areas assigned to the re- 2.0 Methodology spective article was selected or, in case of equality, the one that leads the corresponding list. To reach the objectives, an exploratory study was developed combining a mixed methodology of monostrand conversion 3.0 Results design with purposive sampling (Teddlie and Tashakkori 2009). For a sample selection, we searched the main collection The number of articles (eighty-four) in which we found def- of WoS using the expression “ontology OR ontologies,” re- initions for the term ontology corresponds to less than half stricting it to the title field. Since the intention was to have a of the sample. As a side note, all percentage values are shown significant sample of the current year’s production, the col- rounded to units, so eighty-four correspond to 39% of 214 lection was performed in November 2018, using the filters (see Figure 1). Of these eighty-four, only five articles do not “year=2018” and “document type=article.” present a contextual approach to information systems. The procedure described above resulted in 477 records These five articles represent a residual percentage (11%) of considered for investigation as the study population. Accord- the number of works in the sample whose approach to the ing to probabilistic techniques, given the size of the selected term ontology is related to their philosophical origin (47 = population, a sample of 214 articles would represent a confi- 100%). In contrast, in articles where the approach is made dence level of 95% and a margin of error of ±5%. Since we in the context of information systems, the difference is aimed to collect the most recent articles and not a random much smaller—definitions were found in seventy-nine sample, the articles were sorted chronologically in descending (47%) of 167 (100%) articles with this approach. order, and the first 214 were selected. In the distribution of the sample articles across the five In the qualitative analysis, contingency analysis was initially broad categories of WoS (see Figure 2), we found a relation- used and, later, the categorical technique in the processing ship between these very same categories and the contexts of the context units (Bardin 2011; Krippendorff 2004). used in contingency analysis. The “arts & humanities” and Contingency analysis takes into account the distribution of “technology” categories present the extreme of this relation- elements and their association, as these aspects constitute a ship. All articles in the sample classified in the former cate- significant point for interpretation while they also provide gory address the term ontology in a philosophical context, context. The place or section of the text in which the sub- while in the latter all papers present an approach in the con- jects appear and their co-occurrence with other topics pro- text of information systems. vide relevant indicators for interpretation, which may be as- Regarding the research areas included in the WoS catego- sociative, equivalent or opposite (Bardin 2011). So, the first ries, we found that the three with the highest number of as- step served to collect the definitions explained in the articles signments are related to the computing field: “computer sci- that make up the sample and, by analyzing the context in ence,” “engineering” and “mathematical & computational which the term ontology occurred, it simultaneously al- biology” (see Table 1). lowed the articles to be classified according to the context in The “philosophy” area appears only in the eighth posi- which the term is addressed (philosophical or information tion, with nine assignments, and we found no definitions systems contexts). In the second phase, we proceeded with for the term ontology in any of these articles. The five arti- clustering according to a propositional-semantic distinc- cles, whose approach is related to the philosophical context, tion, delineating categories according to the propositional 202 Knowl. Org. 47(2020)No.3 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

Figure 1. Sample distribution according to the results of contingency analysis.

Figure 2. Sample distribution according to the five major categories of Web of Science. in which definitions were found, are classified in the follow- with definitions. Of this total number of authors only those ing areas: “cultural studies,” “education & educational re- who are cited more than once are presented in Table 2. In search,” “ethnic studies,” “history” and “sociology.” In po- this table, we highlight the difference between the most sitions prior to the one occupied by the “philosophy” area, cited, namely, Thomas R. Gruber with thirty-four men- three areas related to medicine stand out: “biochemistry & tions, and the second, V. Richard Benjamins and Rudi Stu- molecular biology,” “biotechnology & applied microbiol- der with only eight. In addition, it is also noteworthy that, ogy” and “medical informatics.” out of these seventeen authors, only Barry Smith is also the In the collection of authors and their works cited in the author of one of the 214 articles in the sample. definitions (84 = 100%), we found that about one third (24 Analyzing the citations in terms of the number of differ- = 29%) of such authors had no reference to other works. ent works by each author (see Table 2, column “no. works”), However, given the co-authorship and the presentation of Gruber shares the top spot with Smith with four works each.1 multiple references in various definitions, the number of The second position, with three different works, is also shared authors (127) is higher than the total number of articles by two authors: Nicola Guarino and Steffen Staab.2 Knowl. Org. 47(2020)No.3 203 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

articles where definitions were no. articles Web of Science research areas found not found 98 49 49 computer science 35 19 16 engineering 21 6 15 mathematical & computational biology 18 2 16 biochemistry & molecular biology 12 0 12 mathematics 11 2 9 biotechnology & applied microbiology 11 8 3 medical informatics 9 0 9 philosophy 7 0 7 science & technology - other topics 6 3 3 chemistry 6 3 3 education & educational research 6 3 3 environmental sciences & ecology 5 4 1 information science & library science 4 1 3 genetics & heredity 4 3 1 geography 4 1 3 social sciences - other topics 4 4 0 operations research & management science

Table 1. Web of Science research areas assigned to four or more sampled articles.

no. citations authors no. works in co-authorship 34 Gruber, T.R. 4 1 8 Benjamins, V.R. 2 2

8 Studer, R. 2 2

6 Fensel, D. 1 1

5 Staab, S. 3 3

4 Guarino, N. 3 2

4 McGuinness, D.L. 2 2

4 Noy, N.F. 2 2

4 Smith, B. 4 2

3 Borst, W.N. 1 0

2 Chandrasekaran, J. & Josephson, J.R. 1 1 (a)

2 Maedche, A 2 1

2 Oberle, D. 1 1 2 Uschold, M. & Gruninger, M. 1 1 (a) 2 Wand, Y. & Weber, R. 2 2 (a)

Table 2. Authors with two or more citations in the collected definitions; (a) authors always cited together in sample articles.

In accounting for citations by work, Gruber once again two or more citations from the total of sixty-one papers stands out with the authorship of the two most cited works cited. (see Table 3). This is expected given the numerical difference As a context for the analysis of the definitions taken in in citations that Gruber presents compared to all other au- the sample (presented in appendices A, B and C) we also col- thors. As in Table 2, Table 3 presents only the works with lect the definitions present in the works listed in Table 3 (see 204 Knowl. Org. 47(2020)No.3 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

ref. no. citations works authors edition year A 20 A translation approach to portable ontology specifications Gruber 1993 Toward principles for the design of ontologies used for 1993 / 1995 B 11 Gruber knowledge sharing (a) C 6 Knowledge Engineering: Principles and Methods Studer; Benjamins & Fensel 1998 Ontology development 101: a guide to creating your first on- D 3 Noy & McGuinness 2001 tology Construction of engineering ontologies for knowledge sharing E 3 Borst 1997 and reuse F 2 Ontologies: Principles, Methods and Applications Uschold & Gruninger 1996 Chandrasekaran; Josephson G 2 What are ontologies, and why do we need them? 1999 & Benjamins H 2 Handbook on ontologies Staab; Studer 2009 (b) I 2 What is an ontology? Guarino; Oberle & Staab 2009 (b)

Table 3. Works with two or more citations in the collected definitions; (a) quotations from this paper refer to both versions of the article, the 1995 version is an amplified revision of a workshop paper presented in 1993; (b) The paper with the ref. I is a chapter from the work with the ref. H.

Table 4). The chronological presentation of Table 4 was se- number of definitions without reference to other works. lected to facilitate the perception of the link between the The situation described relates to the fact that many of the various works. It seems that the works following Gruber’s definitions with direct or indirect citation refer to the works articles incorporate his definition in a more or less explicit presented in Table 4 (particularly Gruber’s two papers) and manner. the categories mirror the definitions contained therein. Ta- We identified the definitions given in texts with the refs. ble 6 presents the distribution, by the four categories of C and E are explicitly due to Gruber’s work (texts with refs. analysis, of direct and indirect citations for the works pre- A and B). In the definition provided in works due to ref. G, sented in Table 3, according to the relationship mentioned despite the fact that “representation vocabulary” is a notion above among the various definitions. From these relation- relevant to defining ontologies, the “conceptualization” no- ships, we form four groups: the definitions of Gruber’s tion still receives more emphasis because for Gruber the works (ref. A and B), those explicitly constructed from these conceptualization is the ontology itself. In turn, in the (ref. C and E), those that refer to “a conceptualization” less works with refs. D, H and I, the central element of the defi- explicitly (ref. F and G) and those that underline the “set of nitions is the “set of concepts.” This, namely the “set of con- concepts” (ref. D, H and I). cepts,” also appears in the definition of work with ref. F; It should be noted that all works included in the “other however, it appears again as “a conceptualization.” works” column of Table 6 are cited only once. This high- As for the definitions collected from the sample, they lights the prevalence of Gruber’s definition, and others in- were organized into three groups. appendix A depicts sev- fluenced by it, as a reference, particularly in the first cate- enty-nine definitions, which we found in the articles ap- gory, where more than half of the definitions (62%) cite this proaching ontology in the context of information systems. author. In contrast, only one citation of category iv (“a ter- In this appendix, we also pointed out twelve definitions that minological artifact”), refers to works included in Table 3. are framed with a reference to the philosophical origin of the The author of this definition (ref. 097), while citing two term ontology. Appendix B reproduces the six definitions works by Gruber, does not use Gruber’s definition in his associated with that same origin but in articles whose ap- own definition of ontology. Another definition included in proach is made in the context of information systems. Fi- category iv deserves mention, the definition ref. 022 which, nally, we present in appendix C the five definitions of the despite making a direct quote, exchanges the original “it de- term ontology found in articles whose approach is made in scribes the concepts” (Antoniou and Kehagias 2000, 623) the context of its philosophical origin. for “it describes the constructs” (Gelbard et al. 2018, 2). From the analysis of the definitions associated with the Crossing the categories of analysis with the broad areas information systems context, four categories emerged: i) a of WoS shows that the category of analysis (i) “a conceptu- conceptualization; ii) a set of concepts; iii) a conceptual alization,” does not include any definition taken from arti- model; and, iv) a terminological artifact (see Table 5). cles classified in the “life sciences & biomedicine” area con- Comparing the four categories, noticeable is the percent- trasting with the “technology” area where the majority is age increase in relation to the total of each category of the distributed in the first two categories (see Table 7). Knowl. Org. 47(2020)No.3 205 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

works definitions An ontology is an explicit specification of a conceptualization. The term is borrowed from philosophy, A: Gruber (1993) and where an ontology is a systematic account of Existence. For knowledge-based systems, what “exists” is ex- B: Gruber (1995) (a) actly that which can be represented. When the knowledge of a domain is represented in a declarative for- malism, the set of objects that can be represented is called the universe of discourse. “Ontology” is the term used to refer to the shared understanding of some domain of interest which may be used as a unifying framework to solve the above problems in the above-described manner. An ontology F: Uschold et al. (1996) necessarily entails or embodies some sort of world view with respect to a given domain. The world view is often conceived as a set of concepts (e.g. entities, attributes, processes), their definitions and their interrelationships; this is referred to as a conceptualisation. In philosophy, the word ontology means a theory about the nature of being, or the kinds of existence … For AI the main question is not what the nature of being is, but what an AI system has to reason about to be able to perform a useful task … Most researchers generally agree on the definition of Gruber, but find E: Borst (1997) it too broad … We will therefore give a definition of ontologies that suits us the best and continue this section with explaining how to make a good ontology: An ontology is a formal specification of a shared conceptualization. Originally, the term “ontology” comes from philosophy—it goes as far back as Aristotle's attempt to classify the things in the world—where it is employed to describe the existence of beings in the world … Many C: Studer et al. (1998) definitions of ontologies have been given in the last decade, but one that characterises best, in our opinion, the essence of an ontology is based on the related definitions in ([Borst, 1997; Gruber, 1993]): An ontology is a formal, explicit specification of a shared conceptualisation. In philosophy, ontology is the study of the kinds of things that exist … In AI, the term ontology has largely come to mean one of two related things. First of all, ontology is a representation vocabulary, often G: Chan- drasekaran et al. specialized to some domain or subject matter. More precisely, it is not the vocabulary as such that quali- (1999) fies as an ontology, but the conceptualizations that the terms in the vocabulary are intended to capture ... In its second sense, the term ontology is sometimes used to refer to a body of knowledge describing some domain, typically a commonsense knowledge domain, using a representation vocabulary. For the purposes of this guide an ontology is a formal explicit description of concepts in a domain of discourse (classes (sometimes called concepts)), properties of each concept describing various features and D: Noy et al. (2001) attributes of the concept (slots (sometimes called roles or properties)), and restrictions on slots (facets (sometimes called role restrictions)). In the first case, we refer to a philosophical discipline, namely the branch of philosophy which deals with the nature and structure of “reality” ... In the second case, which reflects the most prevalent use in Com- H: Staab et al. (2009) and puter Science, we refer to an ontology as a special kind of information object or computational artifact … I: Guarino et al. (2009) (b) The backbone of an ontology consists of a generalization / specialization hierarchy of concepts, i.e., a taxonomy.

Table 4. Works with two or more citations in the collected definitions; (a) the definition in both works is the same; (b) quotations that cite the work H refer to the definition presented in the work I.

categories ref. of definitions in appendix A dir ind self total 006; 014; 016; 018; 027; 032; 033; 037; 061; 070; 075; 076; 081; 106; 061; 5 16 2 23 i) a conceptualization 125; 131; 143; 152; 154; 178; 195; 210. (22%) (70%) (8%) (100%) 005; 007; 008; 020; 021; 044; 063; 066; 072; 089; 096; 102; 105; 107; 118; 1 20 10 31 ii) a set of concepts 129; 136; 137; 139; 151; 161; 172; 175; 177; 179; 182; 183; 197; 206; 212; (3%) (65%) (32%) (100%) 214. iii) a conceptual 1 3 5 9 004; 025; 068; 079; 080; 126; 132; 142; 180. model (11%) (33%) (56%) (100%) iv) a terminological 015; 022; 023; 036; 047; 050; 069; 085; 097; 104; 122; 148; 157; 170; 193; 1 6 9 16 artifact 213. (7%) (38%) (56%) (100%)

Table 5. Categories resulting from the analysis of the definitions associated with the information systems context, respective definitions and types of quotations; (“dir”—count of definitions with direct quotation; “ind”—count of definitions with indirect quotation; “self”—count of definitions without citing other documents).

206 Knowl. Org. 47(2020)No.3 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

categories of analysis A+B C+E F+G D+H+I other works total i) a conceptualization 15 (62%) 3 (11%) 0 (0%) 2 (8%) 5 (19%) 26 (100%) ii) a set of concepts 5 (19%) 2 (7%) 2 (7%) 3 (11%) 15 (56%) 27 (100%) iii) a conceptual model 2 (29%) 1 (14%) 0 (0%) 0 (0%) 4 (57%) 7 (100%) iv) a terminological artifact 1 (10%) 0 (0%) 0 (0%) 0 (0%) 9 (90%) 10 (100%)

Table 6. Distributions of direct and indirect citations to works presented in Table 3 by the categories resulting from the analysis of the definitions associated with the information systems context; (the total reflects the sum of works cited associated to definitions in appendix A and not the sum of those definitions: “A+B”—sum of citations to Gruber (1993) and Gruber (1995); “C+E”—sum of citations to Studer et al. (1998) and Borst (1997); “F+G”—sum of citations to: Uschold et al. (1996) and Chandrasekaran et al. (1999); “D+H+I”— sum of citations to Noy et al. (2001), Staab et al. (2009) and Guarino et al. (2009).)

categories of analysis WoS major categories total i) ii) iii) iv) Life Sciences & Biomedicine 0 (0%) 8 (47%) 1 (6%) 8 (47%) 17 (100%) Physical Sciences 1 (33%) 2 (67%) 0 (0%) 0 (0%) 3 (100%) Social Sciences 2 (33%) 2 (33%) 0 (0%) 2 (33%) 6 (100%) Technology 21 (40%) 19 (36%) 8 (15%) 5 (9%) 53 (100%)

Table 7. Distributions of the definitions associated with the information systems context according to the major categories of Web of Science; (the WoS category “arts & humanities” was not presented, because it does not include definitions associated with the information systems context; categories of analysis: i) - a conceptualization; ii) - a set of concepts; iii) - a conceptual model; iv) - a terminological artifact).

It is worth noticing that two in five definitions in the WoS formation systems and 11% in that associated with philo- category “technology,” included in our category of analysis sophical origin). Such an occurrence will be related to the (iv), come from the library and information science (LIS) novelty of the connection of the term ontology to digital in- area. Indeed, since the other two definitions of this area fall formation systems compared to the secular association with into the category of analysis (iii), the four definitions of LIS the philosophy area. do not contribute to the majority mentioned above. The Another aspect that promotes the need to present an ex- majority of definitions distribution by categories of analysis plicit definition of a term, besides its novelty, is its use in a (i) and (ii) within the WoS major category “technology.” different context. The use of the term ontology in discipli- Finally, concerning the definitions associated with the nary contexts other than philosophy is one such case. In the philosophical context (presented in appendices B and C), sample, the definitions for the term ontology were all found they present a certain uniformity that can be summarized in in articles outside its original disciplinary context, so the the following statement: “ontology is the philosophical sample seems to mirror the situation described. In the new study of what exists in reality.” Note that, as we mention it disciplinary context, by restricting the count to articles clas- above, none of those definitions are included in articles classified in the WoS “computer science” area, the equal num- sified, on WoS, in the “philosophy” area. ber of articles with and without definition is a potential indicator of the assimilation of the term ontology by the re- 4.0 Discussion spective community. One should expect that as the term becomes more commonly used in a new context, the presenta- In view of the differences in the number of actors in each tion of a definition for it will decrease. field of investigation and the representation of these areas In the conceptual accommodation phase of a term in a in WoS, the comparison of values between them should be new context, the use of sources of recognized authority at- understood in the context of the sample. Despite this limi- testing its meaning is expected. Gruber’s articles (1993; tation, we consider that the sample collected is a strong in- 1995) seem to fulfil this role in research related to compu- dicator of the current prevalence of the information tech- tational ontologies even because of his early association with nology area in studies related to ontology. This superiority it. This role is reflected not only in the corpus of analysis but per se would be sufficient to justify a difference between the also in the works of other authors that these articles cite. We number of definitions found in each context analyzed, but find several references to him in the works cited. not as pronounced as that found (47% in the context of in- Knowl. Org. 47(2020)No.3 207 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

Another possible indicator of Gruber’s role as an “au- come from Barry Smith’s well-documented position on this thority” is his citation in definitions proposed by others. Ex- (e.g., Klein and Smith 2010; Smith 2004; Smith 2006; emplary cases of this appropriation are the definitions ref. Smith, Ceusters and Temmerman 2005) and his action as a 178 and 210, which correspond to that of Studer et al. member of the outreach working group of the Open Bio- (1998), and ref. 076, which is the one presented by Borst logical and Biomedical Ontology Foundry. (1997); all three are referenced in the respective sample arti- Regarding the area of library and information science, cles with Gruber’s work (1993). The case of definition ref. the small number of articles limits possible inferences (only 210 is the most striking because it comes as a direct quote. five articles were accounted, see Table 1). In future research, Although the close connection between the two definitions we consider it relevant to verify whether this small number (Studer’s and Borst’s) and Gruber’s is conducive to confu- is a reflection of the low expressiveness of the area in relation sion (see Table 4), greater rigor would be expected in the to the ontological study in the context of information sys- context of scientific writing. tems, as the sample seems to indicate. Almeida (2013) sug- There are, also, other occurrences that point to the ap- gests that, although ontological study has been present in parent “status” achieved by Gruber’s definition. The defi- the area since the nineteenth century for the representation nition in the expanded version, “an ontology is a formal, ex- of subjects, the LIS literature will be at an early stage with plicit specification of a shared conceptualization” as stated regard to the role that its researchers can play in the model- by Studer et al. (1998, 184), is used in three works (ref. 006, ing of computational ontologies. Another issue that the five 014 and 143) without the respective article being cited. In LIS works in the sample raise concerns concerning their its place, are referenced three other works that define an on- countries of origin, given the absence of the nation consid- tology as: 1) “a special kind of information object or com- ered dominant in ontology research, namely, the USA (Zhu putational artifact” (Staab and Studer 2009, 2); 2) “an engi- et al. 2015). These five works come from Brazil (two), Ger- neering artifact, constituted by a specific vocabulary used to many, India and Sweden. describe a certain reality” (Maedche 2002, 665:11); and, 3) As for the definitions presented for the term ontology in “a specification mechanism to enhance knowledge sharing the philosophical context (depicted in appendix C), some and reuse across different applications” (P. Borst, Akker- considerations are in order. Although the authors of several mans and Top 1997, 365). and diverse fields that propose definitions to ontologies, Still, regarding Gruber’s definition, it is important to these are authors that almost certainly have no research in the mention a question that emerged from this study: the re- field of philosophy per se. This situation is a result of the ap- duced number of clarifications found about what is meant propriation of the term by computer science in the 1980s as by “conceptualization” (only three: ref. 089, 131 and 143). described, for example, by Gruber (1992). Even in the philo- This is an issue that could constitute a study in its own right sophical field, it is considered almost impossible to reach a given the central role of the term “conceptualization” in the consensual definition for the term (Niiniluoto 2002) and respective definitions and the evidence of situations similar even then, the ontological issues are in general referred to as to those mentioned above. The clarification found in the hypotheses (for example, the problem of universals). article with the definition ref. 089 attributes this clarifica- The appropriation of a term from one area of knowledge tion to Guarino et al. (2009) who, in turn, quote directly by another is common practice in the scientific community: from Gruber (1993). Indeed, Guarino only suggested that “emerging technologies require new words and frequently the approach to conceptualization should be intensional, in borrow from other fields which may be contiguous or to- contrast to Gruber’s original perspective. Within this de- tally unrelated” (English 1998, 32). Some consider that such bate, the word “intensional” is used in the formal semantics appropriation related with ontology began in the 1960s, context, where one can find the pair extensional-inten- when Mealy (1967, 525) in a paper about the nature and sional. Extensional sentences make reference to sentences models for data suggested: “we could easily resurrect dis- that depend only on local-facts for their truth-conditions, putes in medieval philosophy at this point! The issue is on- while intensional sentences are those that are not exten- tology, or the question of what exists.” The idea is that in sional (Portner 2005). In ontologies, the intensional part is knowing the nature of entities, in understanding the struc- called T-BOX, the terminological part, which refers to axi- ture of the world provided by an ontology, one can transfer oms about properties and relationships, for example, a hu- part of this structure for the computer. In doing this, one man being is a person; the extensional part is called A-BOX, would model better than by using ad hoc approaches and the assertional part, which refers to instances of classes, for solipsism. To that end, another early use of the term ontol- example, Jonh loves Mary. ogy in computer-aided information systems was the work of One area of research where the view of ontology as a con- Hayes (1983), seeking to provide an adequate theory of the ceptualization does not appear to have much adherence is common-sense world (Smith and Casati 1994). Unlike this “life sciences & biomedicine.” A possible influence may line of research, which retains the notion of philosophy’s 208 Knowl. Org. 47(2020)No.3 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies ontology, the use of the term ontology to designate a vocab- 6.0 Final remarks ulary expressed in a knowledge representation language seems to deviate from the metaphysical sense. This second We consider that the panorama described in this exploratory use of the term appears to undergo a “metaphorizing” pro- study, despite its limitations, presents empirical data that cess. As stated by Rita Temmerman (1995, 125): “metapho- contribute to the understanding of the lack of clarity re- rising is the result of encoding at the concept level. The regarding the meaning of the term ontology in the context of sulting name or term for the concept can be understood in information systems. In particular, we have uncovered the its new meaning without understanding the basis for the use of the term ontology as a synonym for conceptualization naming.” In this process, the new meaning may deviate and its consequent indistinction regarding other types of from the original, because it is built in a new framework of representational artifacts. An uncritical use of terminology assumptions. makes communication difficult, leading to misunderstand- ings. Also, LIS cannot properly contribute with its expertise 5.0 Conclusion in classification, if terminological difference hampers the initiatives. Within this situation it is difficult to distinguish This study points to evidence that leads us to consider that between the specificities of a reference ontology, coming the term ontology is being consolidated in the context of in- from a scientific domain, in comparison to those of another formation systems with a different meaning from the origi- system, also called ontology, designed to meet the generally nal. The emphasis in most definitions is placed on the no- idiosyncratic needs of a particular institution or particular tions of: a) “a conceptualization;” or, b) “a set of concepts,” interests. While in the latter it is the pragmatic aspects that moving away from the ontological study as a branch of phi- fundamentally constrain it, in the former the primacy losophy. The philosophical study focuses between a purely should be the scientific correction of the model. This study formal point of view, too general to contain relevant infor- points out that, for the computing community in particu- mation, and the cognitive point of view, laden with implicit lar, this distinction does not seem to be very relevant. We information (Poli 2010b). It will be in this last pole that the also revelaed a scant citation of works by voices disagreeing ontologies, as conceptualizations, fit. In these, only the par- with that position. We admit that additional in-depth stud- ticular point of view of a community matters, as Gruber ies are necessary to assess the consistency of the trend we de- (2004, 5) states: “I find it critical to remember that every on- tected, as well as a greater sample that allows better represen- tology is a treaty—a social agreement—among people with tation of research areas. It is worth mentioning that such an some common motive in sharing.” To the extent that model enlargement would apply in an extension of the temporal correction seems less important than its usefulness: “we horizon of our search and, which would not be representa- don't have to worry so much about whether they are right tive of a single year’s production. In any case, we hope to and getting on with the business of building them to do have made a reflective contribution in a subject where a something useful” (Gruber 2004, 1). careful intervention can be the distinguishing mark of the Parallel to Gruber’s understanding of ontology in infor- KO and LIS areas. mation systems, some authors have presented definitions with a description that can be considered closer to their use Notes in the context of philosophy. These definitions have in common a deviation from the cognitive load inherent in the 1. Gruber’s works, besides the two presented in Table 2, term “concept,” using other terms, such as “entities,” to des- are: (Gruber 2009; Gruber and Olsen 1994); and Smith’s ignate the elements to be represented in an ontology. De- works are: (Arp, Smith and Spear 2015; Munn and spite this possible link between the two research areas, the Smith 2008; Smith 2003). study points to the lack of interaction between them. Ex- 2. Staab’s work, besides the two presented in Table 2, is: trapolating the tendency, we could foresee that the meaning (Maedche and Staab 2001); and Guarino’s works, be- of the term ontology, and even possibly the study itself, in sides the one presented in Table 2, are: (Guarino 1998; the context of information systems will have only a remotely Guarino and Giaretta 1995). historical connection with its counterpart in the field of philosophy. In this context, we consider the position of au- References thors such as Poli or Smith, who regard ontological study as interdisciplinary, crucial for the continuation of a healthy Almeida, Maurício Barcellos. 2013. “Revisiting Ontologies: and useful discussion between the two fields. A Necessary Clarification.” Journal of the American So- ciety for Information Science and Technology 64: 1682-93. doi:10.1002/asi.22861 Knowl. Org. 47(2020)No.3 209 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

Almeida, Maurício Barcellos and Marcello P. Bax. 2003. tin—AIS Special Interest Group on Semantic Web and In- “Uma visão geral sobre ontologias: Pesquisa sobre formation Systems 1, no. 3: 1-5. definições, tipos, aplicações, métodos de avaliação e de Gruber, Thomas R. 2009. “Ontology.” In Encyclopedia of construção.” Ciência Da Informação 32, no. 3: 7-20. Database Systems I, ed. Ling Liu and M. Tamer Özsu. doi:10.1590/S0100-19652003000300002 New York: Springer, 1963-5. Antoniou, Grigoris and Athanasios Kehagias. 2000. “On Gruber, Thomas R. and Greg R. Olsen. 1994. “An Ontology the Refinement of Ontologies.” International Journal of for Engineering Mathematics.” In Principles of Knowledge Intelligent Systems 15. Representation and Reasoning: Proceedings of the Fourth Arp, Robert, Barry Smith and Andrew D. Spear. 2015. International Conference KR'9, ed. Jon Doyle, Piero Building Ontologies with Basic Formal Ontology. Cam- Torasso, & Erik Sandewall. San Francisco: Morgan Kauf- bridge, MA: MIT Press. mann, 258-69. Bardin, Laurence. 2011. Análise de conteúdo, trans. Luís Gruninger, Michael, Olivier Bodenreider, Frank Olken, Antero Reto and Augusto Pinheiro. São Paulo: Leo Obrst and Peter Yim. 2008. “Ontology Summit Almedina. Translation of L'analyse de contenu, 1977. 2007 - Ontology, Taxonomy, Folksonomy: Understand- Borst, Pim, Hans Akkermans and Jan Top. 1997. “Engineer- ing the Distinctions.” Applied Ontology 3: 191-200. ing Ontologies.” International Journal of Human-Com- doi:10.3233/AO-2008-0052 puter Studies 46: 365-406. doi:10.1006/ijhc.1996.0096 Guarino, Nicola. 1998. “Formal Ontology in Information Borst, Willem Nico. 1997. “Construction of Engineering Systems.” In Formal Ontology in Information Systems: Ontologies for Knowledge Sharing and Reuse.” PhD diss., Proceedings of the first International Conference (fois'98), Universiteit Twente. https://ris.utwente.nl/ws/portal- June 6-8, Trento, Italy, ed. Nicola Guarino. Amsterdam: files/portal/6036651/t0000004.pdf IOS Press, 3-15. Chandrasekaran, B., J.R. Josephson and V.R. Benjamins. Guarino, Nicola and Pierdaniele Giaretta. 1995. “Ontolo- 1999. “What Are Ontologies and Why Do We Need gies and Knowledge Bases: Towards a Terminological Them?” IEEE Intelligent Systems 14: 20-6. doi:10.1109/ Clarification.” In Towards Very Large Knowledge Bases. 5254.747902 Knowledge Building and Knowledge Sharing, ed. Nico- Currás, Emilia. 2010. Ontologies, Taxonomies and Thesauri laas J. I. Mars. Amsterdam: IOS Press, 25-32. in Systems Science and Systematics. Oxford: Chandos Guarino, Nicola, Daniel Oberle and Steffen Staab. 2009. Publishing. doi:10.1533/9781780631752 “What Is an Ontology?” In Handbook on Ontologies, ed. Dahlberg, Ingetraut. 2014. “What Is Knowledge Organiza- Steffen Staab and Rudi Studer. 2nd ed.. Berlin: Springer, tion?” Knowledge Organization 41: 85-91. 1-17. doi:10.1007/9783540926733 English, Kathryn. 1998. “Understanding Science: When Guizzardi, Giancarlo. 2007. “On Ontology, Ontologies, Metaphors Become Terms.” Asp, nos. 19/22: 151-63. Conceptualizations, Modeling Languages and (Meta) doi:10.4000/asp.2800 Models.” In Databases and Information Systems IV, ed. Gelbard, Roy, Roni Ramon-Gonen, Abraham Carmeli, Olegas Vasilecas, Johann Eder and Albertas Caplinskas. Ran M. Bittmann and Roman Talyansky. 2018. “Senti- Amsterdam: IOS Press, 18-39. ment Analysis in Organizational Work: Towards an On- Hayes, Patrick J. 1983. “The Second Naive Physics Mani- tology of People Analytics.” Expert Systems 35, no. 5: festo.” Cognitive Science Technical Report URCS-10. e12289. doi:10.1111/exsy.12289 Rochester, NY: University of Rochester. Gilchrist, Alan. 2003. “Thesauri, Taxonomies and Ontolo- Hennig, Boris. 2008. “What Is Formal Ontology?” In Ap- gies – an Etymological Note.” Journal of Documentation plied Ontology: An Introduction, ed. Katherine Munn 59: 7-18. doi:10.1108/00220410310457984 and Barry Smith. Heusenstamm: Ontos Verlag, 39-56. Gruber, Thomas R. 1992. “What Is an Ontology?” Inter- Herre, Heinrich. 2015. “Formal Ontology: A New Disci- national Journal Human-Computer Studies 43: 907-28. pline Between Philosophy, Formal Logic, and Artificial Gruber, Thomas R. 1993. “A Translation Approach to Intelligence.” In Kybernetik, Logik, Semiotik. Philoso- Portable Ontology Specifications.” Knowledge Acquisi- phische Sichtweisen, ed. Fuchs-Kittowski Klaus, Rainer E. tion 5: 199-220. Zimmermann, Wolfgang Coy, Manfred Bierwisch, Gruber, Thomas R. 1995. “Toward Principles for the De- Heinrich Herre and Eckardt Michael. Berlin: Trafo Wis- sign of Ontologies Used for Knowledge Sharing.” Inter- senschaftsverlag. national Journal Human-Computer Studies 43: 907-28. Hjørland, Birger. 2019. “Classification.” Knowledge Organ- doi:10.1006/ijhc.1995.1081 ization 44, no. 2: 97-128. Gruber, Thomas R. 2004. “Every Ontology Is a Treaty–a Khazraee, Emad and Xia Lin. 2011. “Demystifying Ontol- Social Agreement–among People with Some Common ogy.” In Classification & Ontology: Formal Approaches Motive in Sharing.” Interview by Miltiadis Lytras. Bulle- and Access to Knowledge: Proceedings of the International 210 Knowl. Org. 47(2020)No.3 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

UDC Seminar, ed. Aida Slavic and Edgardo Cival- Computer Applications, ed. Roberto Poli, Michael Healy lero.Würzburg: Ergo, 41-54. and Achilles Kameas. Dordrecht: Springer Netherlands, Klein, Gunnar O. and Barry Smith. 2010. “Concept Sys- 1-26. doi:10.1007/978-90-481-8847-5_1 tems and Ontologies: Recommendations for Basic Ter- Portner, Paul H. 2005. What is Meaning? Fundamentals of minology.” Transactions of the Japanese Society for Artifi- Formal Semantics. Oxford: Blackwell. cial Intelligence 25: 433-41. doi:10.1527/tjsai.25.433 Smiraglia, Richard P. 2014. “Ontology.” In The Elements of Kless, Daniel, Jutta Lindenthal, Simon Milton and Edmund Knowledge Organization. Cham: Springer International Kazmierczak. 2011. “Interoperability of Knowledge Or- Publishing, 43-50. ganization Systems with and through Ontologies.” In Smith, Barry. 2003. “Ontology.” In Blackwell Guide to the Classification and Ontology: Formal Approaches and Access Philosophy of Computing and Information, ed. Luciano to Knowledge: Proceedings of the International UDC Sem- Floridi. Oxford: Blackwell, 155-66. inar 19 - 20 September 2011, The Hague, The Netherlands, Smith, Barry. 2004. “Beyond Concepts: Ontology as Real- ed. Aida Slavic and Edgardo Civallero. Würzburg: Ergon, ity Representation.” In Formal Ontology in Information 55-74. Systems: Proceedings of the Third International Confer- Krippendorff, Klaus. 2004. Content Analysis: An Introduc- ence (FOIS 2004), ed. Achille C. Varzi and Laure Vieu. tion to Its Methodology. 2nd ed. Thousand Oaks, CA: Sage Frontiers Iin Artificial Intelligence Aand Applications Publications. 114. Amsterdam: IOS Press, 73-84. Maedche, Alexander. 2002. Ontology Learning for the Se- Smith, Barry. 2006. “From Concepts to Clinical Reality: An mantic Web. The Kluwer International Series in Engi- Essay on the Benchmarking of Biomedical Terminolo- neering and Computer Science 665. Boston, MA: gies.” Journal of Biomedical Informatics 39: 288-98. Springer. doi:10.1007/978-1-4615-0925-7 doi:10.1016/j.jbi.2005.09.005 Maedche, Alexander and Steffen Staab. 2001. “Ontology Smith, Barry. 2013. “Classifying Processes: An Essay in Ap- Learning for the Semantic Web.” IEEE Intelligent Sys- plied Ontology.” In Classifying Reality, ed. David S. tems 16: 72-9. doi:10.1109/5254.920602 Oderberg. Chichester, UK: John Wiley, 101-26. Mealy, George H. 1967. “Another Look at Data.” In AFIPS doi:10.1002/9781118627747.ch6 '67 (Fall): Proceedings of the November 14-16, 1967, Fall Smith, Barry and Roberto Casati. 1994. “Naive Physics: An Joint Computer Conference. New York: ACM Press, 525- Essay in Ontology.” Philosophical Psychology 7: 225-44. 34. doi:10.1145/1465611.1465682 Smith, Barry, Werner Ceusters and Rita Temmerman. 2005. Munn, Katherine and Barry Smith, eds. 2008. Applied On- “Wüsteria.” Studies in Health Technology and Informat- tology: An Introduction. Heusenstamm: Ontos Verlag. ics 116: 647-52. Niiniluoto, Ilkka. 2002. Critical Scientific Realism. 2nd ed. Smith, Barry, Waclaw Kusnierczyk, Daniel Schober and Oxford: Oxford University Press. Werner Ceusters. 2006. “Towards a Reference Terminol- Noy, Natalya F. and Deborah L. McGuinness. 2001. “On- ogy for Ontology Research and Development in the Bi- tology Development 101: A Guide to Creating Your omedical Domain.” In KR-MED 2006 Proceedings: Sec- First Ontology.” https://protege.stanford.edu/publica- ond International Workshop on Formal Biomedical tions/ontology_development/ontology101.pdf Knowledge Representation Baltimore, Maryland No- Poli, Roberto. 1996. “Ontology for Knowledge Organiza- vember 8, 2006, ed. Olivier Bodenreider. CEUR Work- tion.” In Knowledge Organization And Change: Proceed- shop Proceedings 222: 57-65. http://ceur-ws.org/Vol- ings of the Fourth International ISKO Conference, Wash- 222/krmed2006-p07.pdf ington, DC, USA, July 15-18, 1996, ed. R. Green. Ad- Soergel, Dagobert. 1999. “The Rise of Ontologies or the Re- vances in Knowledge Organization 5. Frankfurt/ invention of Classification.” Journal of the American Soci- Main: Indeks, 313-9. ety for Information Science and Technology 50: 1119-20. Poli, Roberto. 2010a. “Preface.” In Theory and Applications Souza, Renato Rocha, Douglas Tudhope and Maurício of Ontology: Computer Applications, ed. Roberto Poli, Mi- Barcellos Almeida. 2012. “Towards a Taxonomy of KOS: chael Healy and Achilles Kameas. Dordrecht: Springer, v- Dimensions for Classifying Knowledge Organization viii. Systems.” Knowledge Organization 39: 179-92. Poli, Roberto. 2010b. “Ontology: The Categorial Stance.” Staab, Steffen and Rudi Studer, eds. 2009. Handbook on In Theory and Applications of Ontology: Philosophical Ontologies. 2nd ed. Berlin: Springer. doi:10.1007/978-3- Perspectives, ed. Roberto Poli and Johanna Seibt. Dor- 540-92673-3 drecht: Springer, 1-22. Studer, Rudi, V. Richard Benjamins and Dieter Fensel. Poli, Roberto and Leo Obrst. 2010. “The Interplay Be- 1998. “Knowledge Engineering: Principles and Meth- tween Ontology as Categorial Analysis and Ontology as ods.” Data & Knowledge Engineering 25: 161-97. Technology.” In Theory and Applications of Ontology: doi:10.1.1.110.8406 Knowl. Org. 47(2020)No.3 211 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

Teddlie, Charles and Abbas Tashakkori. 2009. Foundations Vickery, B. C. 1997. “Ontologies.” Journal of Information of Mixed Methods Research: Integrating Quantitative Science 23: 277-86. doi:10.1177/016555159702300402 and Qualitative Approaches in the Social and Behavioral Zhu, Qiaoli, Xuesong Kong, Song Hong, Junli Li and Sciences. Los Angeles: SAGE Publications. Zongyi He. 2015. “Global Ontology Research Progress: Temmerman, Rita. 1995. “The Process of Revitalisation of A Bibliometric Analysis.” Aslib Journal of Information Old Words: ‘Splicing’ a Case Study in the Extension of Management 67: 27-54. doi:10.1108/AJIM-05-2014- Reference.” Terminology 2: 107-28. doi:10.1075/term.2. 0061 1.06tem Uschold, Michael and Michael Grüninger. 1996. “Ontolo- gies: Principles, Methods and Applications.” Knowledge Engineering Review 11: 93-136.

Appendices

We present the definitions found in the sample in appen- documents are identified with “dir,” in the case of direct dices A, B and C and in D the references of the respective quotations, and with “ind,” when quotation is indirect. In articles. In transcribing the definitions with citation (direct appendix A, the designations “self,” “ind” and “dir,” may or indirect), the original numerical references were replaced have the suffix “_ro” when the definition is framed with a by the author-date style. On the column “WoS cat.” in ap- reference to the origin of the term ontology (which may not pendices A, B and C, we present the Web of Science major be included in the transcribed excerpt) or “_pd” when the category where the respective article was classified. In the authors also define the term in that context of origin (those “type” column we identified definitions that do not refer to definitions are in appendix B). other work with the designation “self,” those citing other

Appendix A: Definitions of the term ontology associated with information systems found in articles whose approach is made in the context of these systems.

Ref. Definition Type WoS Cat. One method of conceptualising data is as an ontology—a knowledge model that formalises var- self Life Sciences & 004 iables, properties, and relationships such that they can be used for problem solving. Biomedicine Swartout and Tate (1999) define ontology as a basic structure or framework around which a ind Technology 005 knowledge-base can be built. Formally, the ontology of a particular domain covers its terminology (domain vocabulary), all essential concepts and their instances (individuals) in the domain. From a computer science point of view and in the context of knowledge acquisition, an ontol- dir Technology 006 ogy could be defined as “a formal, explicit specification of a shared conceptualisation” (Staab et al., 2009). 007 Ontology is a formal representation of a set of domain concepts and their relationships. self Technology An ontology can be used to describe concepts, attributes, and restrictions within a particular ind Technology 008 domain [Noy & McGuinness, 2001]. Ontologies are formal, explicit specifications of a shared conceptualization to represent domain ind Technology 014 knowledge, [Maedche, 2012] which are widely applied recently to represent knowledge in the Semantic Web. Applied ontologies consist of a set of clearly defined entities (which may, however, each have self Life Sciences & 015 multiple labels), structured hierarchically, and interconnected by defined relations. Biomedicine Ontology engineering is a powerful tool for computer-based information modeling and man- ind Technology 016 agement that aims to conceptualize the physical world in a formal and explicit manner (Gruber, 1993). Ontology is defined as a formal representation of knowledge pertaining to a particular domain ind Technology [Gruber, 1993; Noy & McGuinness, 2001] .... Ontologies are crucial to enable the vision of se- 018 mantic web. They provide a common and shared understanding of concepts in a specific domain, allow reuse of domain knowledge, and make the data interoperable. Ontology is a formal model that represents a target domain which is generally constituted by a ind _pd Technology 020 hierarchy of concepts that are interrelated by defined relations [McGuinness & Van Harmelen, 212 Knowl. Org. 47(2020)No.3 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

Ref. Definition Type WoS Cat. 2004].

[A]n ontology is an explicit formal specification of the terms in the domain and relations ind Life Sciences & among them [Gruber, 1993] expressed in machine-readable language; therefore, they can be Biomedicine 021 processed automatically…. Classes (i.e., concepts), subclasses, and predicates between concepts represent an ontology. According to Antoniou and Kehagias (2000, p. 623), “An ontology defines the terminology of dir Technology 022 a domain: it describes the constructs that constitute the domain, and the relationships between those constructs.” [B]iomedical ontologies are sets of terms and relations that represent biomedical entities and self Life Sciences & 023 how they connect with one another. Biomedicine As a conceptual model, ontology storage and management the information, has been widely self _ro Technology 025 concerned in the field of information retrieval. 027 “Ontology is defined as an explicit specification of a conceptualization” (Gruber, 1993). dir Technology The definition of ontology has many variants, which can be generalized under a machine-read- self Technology 032 able, structured representation of information …. the business domain ontology is intended to specify the conceptualization of a particular real-world business domain. [O]ntology is a conceptualization with explicit specification in unanimity filed knowledge, it self Technology 033 has been used popularly in modeling and retrieval of knowledge engineering for product lifecy- cle management. One way of modeling data is designing ontologies and using them to maximize the benefit of self Life Sciences & 036 accessing and extracting valuable implicit and explicit knowledge from structured and unstruc- Biomedicine tured data. The scientific meaning derives from Information Science, where “ontology” refers to a shared ind _pd Social Sciences 037 conceptualisation of a domain, presented as an organised technical vocabulary for that domain [Gruber, 1993]. Traditionally a philosophical concept, ontology has been adopted by computer science and in- ind _ro Life Sciences & formation science as a new way of defining meaning and relationships within data [Feilmayr & Biomedicine 044 Wöß, 2016; Gruber, 1995].Ontologies are composed of three parts: a set of vocabulary representing various concepts, definitions for the vocabulary set, and defined relationships between the concepts. Ontologies are models of reality, and are expressed through entities and their relationships. self Life Sciences & 047 Biomedicine Ontologies are information artifacts that present two basic characteristics: they are vocabularies ind Life Sciences & 050 shared in a certain community, and they have formal semantics based on axioms, expressed in Biomedicine some logic language. [Horrocks, 2008]. Ontology is defined as an explicit specification of a conceptualization, and is consisted of con- ind Technology 061 cepts, relations, axioms which describe a certain domain [Gruber, 1993]. 063 An ontology is a unanimous agreement on shared concepts. self _ro Social Sciences Simply, an ontology is a machine-readable artifact that encodes a logical representation of a do- self_pd Life Sciences & main space using vocabularies, and their semantic meanings..... In general, knowledge in an on- Biomedicine 066 tology is represented as triple which is information presented in subject>predicate>object. Es- sentially, the subject>predicate>object are concepts. Ontology has been widely applied ... as a model for storing and representing knowledge. Owing self _ro Technology 068 to it is an effective concept semantic model and a powerful analysis tool. Formally, an ontology is “a representational artifact, comprising a taxonomy as proper part, dir Life Sciences & 069 whose representations are intended to designate some combinations of universals, defined clas- Biomedicine ses and certain relations between them” [Smith et al., 2006]. Ontology is defined as an explicit and shared conceptualization of a given domain that provides ind Technology explicit logical assertions about three types of things, classes, instances, and properties, and of- 070 fers the means to capture and convert human knowledge into a computer-understandable and explicit format. [Gruber, 1993; Richards & Simoff, 2001]. The term ontology refers to “a representation and definition of the categories, properties, and dir Life Sciences & 072 relations of the concepts, data, and entities that substantiate one, many, or all domains.” Biomedicine [Gruber, 2009]. Knowl. Org. 47(2020)No.3 213 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

Ref. Definition Type WoS Cat. Ontologies are defined as formal, explicit specifications of shared conceptualisations. [Gruber, ind Technology 075 1993]. 076 An ontology is a formal specification of a shared conceptualization. [Gruber, 1993]. ind Technology Ontology is defined in informatics as an attempt for comprehensive and detailed formalization self _ro Technology 079 of a given area of knowledge via a conceptual scheme. 080 Ontologies provide conceptual models to represent and share knowledge. self Technology The term ontology is used in the literature of computer science to refer to an explicit specifica- ind Technology 081 tion of a shared conceptualization in a machine readable format. [Gruber, 1993; Guarino & Gi- aretta, 1995; Studer; Benjamins & Fensel, 1998]. Ontologies help integrate disparate or unorganized data to produce meaning, sort of “like a the- self Social Sciences 085 saurus, a finite set of terms, organized as a hierarchy that can be used to provide a value for an element. ISO 1087-1:2000 defines “concept” as a “unit of knowledge created by a unique combination self Technology 089 of characteristics.” In general, ontologies thus can be seen as collections of knowledge for some specific domain of discourse. An ontology describes domain knowledge or domain space that represents and connects con- self Life Sciences & cepts of the domain.... The resulting software artifact can then be integrated with other soft- Biomedicine 096 ware components to provide extended capabilities, perform tasks, and enable machine reasoning. An ontology is a computational representation of a domain of knowledge based upon a con- ind Life Sciences & 097 trolled, standardized vocabulary for describing entities and the semantic relationships between Biomedicine them. [David & Tim, 2012; Gruber, 1993; Gruber & Olsen, 1994; Guarino, 1998] [A]n ontology is an explicit formal specification of the concepts (also referred to as classes) in a ind Technology 102 domain and the relations among them. [Uschold & Gruninger, 1996] An ontology can be defined as a set of hierarchically ordered terms to represent a specific do- self Technology 104 main. [original in portuguese, free translation of authors] “An ontology is a formal explicit description of concepts in a domain of discourse (classes dir Technology (sometimes called concepts)), properties of each concept describing various features and attrib- 105 utes of the concept (slots (sometimes called roles or properties)), and restrictions on slots (facets (sometimes called restrictions)).” [Noy & McGuinness, 2001]. In computer science, ontologies can be defined as “explicit specification of a shared conceptual- dir Physical Sciences 106 ization.” [Gruber, 1993]. It is a specialised representation artefact, designed to usefully mediate between an artificial net- self Physical Sciences 107 work of labelled concepts and the perceptions and classifications of the people who work with it. An ontology is an explicit specification of a conceptualization. [Edgington et al., 2004; Gruber, ind Technology 113 1994]. In computer and information sciences, ontologies serve as explicit representations of the con- ind Social Sciences 118 cepts and relationships relevant to some area of interest (Gruber, 1995). Ontology is a description of knowledge about a domain of interest, the core of which is ma- self Technology 122 chine processable specification with a formally defined meaning. Ontology is defined as an explicit specification of a shared conceptualization representing ind Technology 125 knowledge through concepts, relationships, and individuals. [Gruber, 1993]. In terms of information technology, ontologies summarize various types of formal conceptual ind Technology systems that allow an explicit and thus machine-processable assignment of meaning to jointly 126 defined linguistic concepts. [Gruber, 1993; Studer; Benjamins & Fensel, 1998]. [original in german, free translation of authors]. Ontology is a formal representation of a domain of knowledge in terms of concepts and rela- ind Technology 129 tionships to data (attributes) and other concepts. [Gruber, 1993]. An ontology is a formal, explicit specification of a shared conceptualization. [Studer; Benja- ind Social Sciences 131 mins & Fensel, 1998]. Using ontology as a foundation for conceptual modeling builds on the assumption that ontol- ind _pd Technology 132 ogy can help to better understand how the world is constituted (Gruber, 1995; Wand et al., 1999; Wand and Weber, 2006). 214 Knowl. Org. 47(2020)No.3 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

Ref. Definition Type WoS Cat. [O]ntologies (i.e., formal descriptions of domains [Gruber, 1993] using characterizing con- ind Technology 136 cepts, individuals, properties, and relations) remains a non-trivial task. Ontology is a tool for representing knowledge and reasoning that serves the organization of a ind Technology 137 set of concepts in a specific field, as well as the relations between these concepts [Agirre et al., 2000; Faatz & Steinmetz, 2002; Parekh; Gwo & Finin, 2004]. Derived from philosophy, in computer science, we refer to an ontology as a special kind of in- ind _ro Technology formation object or computational artifact [Guarino; Oberle & Staab, 2009; Studer, Benjamins 139 & Fense, 1998].... These ontologies contains concepts and relationships that are meant to describe or record the facts about the real world. Ontology-based data management aims at managing data through the lens of an ontology, that self Technology 142 is, a conceptual representation of the domain of interest in the underlying information system. As it has already been stated, an ontology is a formal explicit specification of a shared conceptu- ind Technology 143 alization of a domain [Borst; Akkermans & Top, 1997]. In informatics, “ontology” refers to these formalized descriptions of the being, of existing enti- ind Social Sciences 148 ties (Smith, 2003). In order to be able to automate engineering activities, a formalization of domain knowledge is self Technology essential. This formalization can be used to model system aspects in sense of properties and rela- 151 tions.... Basically, all parameters, components, and material types are modeled as concepts and individuals in the ontology. Each ontology is the formal specification of the shared conceptualization of a domain of study ind Technology 152 (Gruber, 1995). Similarly for standardization in description of cloud services, it is required to use a semantic ind Technology knowledge representation strategy called ontology that points to a perception of the domain of 154 interest. It incorporates a set of concepts those include entities, attributes and their inter-relationships are altogether referred to as a conceptualization [Gruber. 1995]. A very versatile knowledge management approach is the use of ontologies, information models self Technology consisting of formally defined hierarchies of entity types describing some domain of interest, 157 coupled with well-defined relations between types and axioms expressing fundamental domain knowledge. Ontology is one of the semantic web technologies to represent, exchange and reuse domain self Technology concepts, relations between concepts, and rules.... Domain ontology is a form of representation 161 of concepts, relations and rules in a specific domain so that information in the domain can be well stored, searched and shared. An ontology is a knowledge-based structured system, which consists of a rich, standardized vo- self Life Sciences & 170 cabulary to describe entities and the semantic relationships between them. Biomedicine Since an ontology is a formal fixation of the agreements made by a group of specialists in a cer- ind Technology tain domain about a system of concepts they use, as well as their properties and axioms, each 172 ontology system makes sense only for the group of people who accept these agreements (the social nature of ontologies). [Beniaminov, 2008] Ontologies can be seen as structured vocabularies that explain the relations among their terms self Technology 175 (or classes). They are formed by concepts and relations that can be combined to form more complex class expressions. Researchers in information systems and knowledge-based systems have expanded the definition ind _pd Technology 177 so that the term ontology refers to, not only the vocabulary itself, but also the concepts the vocabulary is intended to express [Chandrasekaran; Josephson & Benjamins, 1999]. Gruber [1993] defines an ontology as a formal explicit specification of a shared conceptualiza- ind Technology 178 tion. According to Borst [1997], ontology can be defined as a formal and explicit specification of a ind Technology 179 set of concepts in a shared form. Consequently, an ontology represents the conceptual model of the specific domain of interest, ind _pd Technology 180 describing it in a declarative fashion [Sonntag, 2010]. An ontology is defined as a specification of concepts and relationships between the concepts ind Technology 182 that can exist in a given setting [Gallardo et al., 2011]. Ontology, as a shared concept, is a mechanism that describes concepts and their system relation- ind Technology 183 ships [Elhdad; Chilamkurti & Torabi, 2013; Verstichel et al., 2011]. Knowl. Org. 47(2020)No.3 215 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

Ref. Definition Type WoS Cat. An ontology can be understood as a logical representation of a domain model [Jones; Bench- ind Technology 193 Capon & Visser, 1998]. An ontology is a formal and explicit specification of a shared conceptualization [Studer; Benja- ind Technology 195 mins & Fensel, 1998], i.e. it provides a formal, structured representation of knowledge, with the advantage of it being reusable and shareable. A proper ontology, however, is defined as a formal representation of knowledge in a certain re- ind Life Sciences & ality (i.e., a certain domain of knowledge), in a way that different people—and, notably, com- Biomedicine 197 puters—can understand the concepts it contains and learn about the reality that is being represented [Arp; Smith & Spear, 2015; Rubin; Shah & Noy, 2007]. The development of a formal system (i.e., ontology) .... Ontologies are used in biology as a way self Life Sciences & 206 to classify terms, how they relate to broader concepts, and their interrelationships…. Formally, Biomedicine concepts are generally called “classes.” According to Gruber (1993), “Ontologies are effectively formal and explicit specifications in dir Technology 210 the form of concepts and relations of shared conceptualizations.” In computer science, an ontology is a taxonomic description of the concepts in an application ind Life Sciences & 212 domain and the relationships among them [Musen, 1998]. Biomedicine In Information Science, ontologies indicate models to represent ontological and epistemologi- ind Technology 213 cal assumptions that are relevant to the understanding of research and its computational treatment (Campos & Campos, 2014). [original in portuguese, free translation of authors] Ontologies can be described as structured vocabularies that explain the relations among their self Physical Sciences 214 terms (or classes). They are formed by concepts and relations that can be combined to form more complex class expressions.

Appendix B: Definitions of the term ontology associated with its philosophical origin found in articles whose approach is made in the context of information systems.

Ref. Definition Type WoS Cat. 020 Philosophically, it is to inquire into being as it is being or into being in so much as they exist. self Technology In Philosophy, the term “Ontology” refers to a discipline investigating what exists most funda- 037 mentally in the real world [Heil, 2003; Varzi, 2011], and hence reflects the most fundamental ind Social Sciences units of thought for theorising about the nature of reality. The word ontology has its roots in metaphysical philosophy, extending back to Aristotle’s Cate- Life Sciences & Bi- 066 (a) gories, as a “nature of being.” omedicine 132 In philosophy, ontology is defined as the study of being, of what there is (Bricker, 2016) ind Technology In philosophy, the study of ontology deals with the nature of reality – exploring the similarities, 177 ind Technology differences and relationships between the types of entities that exist [Davidson, 2013]. The word ontology has borrowed initially from philosophy within less than twenty years, and it 180 self Technology means the philosophical study of nature of existence. (a) Although the definition refers to Aristotle's, it is not formalized in the bibliographic references presented in the respective article.

216 Knowl. Org. 47(2020)No.3 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

Appendix C: Definitions of the term ontology associated with the context of its philosophical origin found in articles whose approach is made in that same context.

Ref. Definition Type WoS Cat. [T]he theory of existence, or more narrowly, of what really exists, as opposed to that which ap- self Social Sciences 046 pears to exist but does not. [O]ntology has historically been used in the field of philosophy to designate a concern with the self Social Sciences 060 existence of things and the essence of being. [T]hus consider ontology as a methodological problem rather than as a specific branch of a self Arts & Humani- 149 philosophical investigation setting up the axiomatic principles of beings or of so‐called histori- ties cal stages. 150 Kohn [2015] defines ontology as “the study of reality.” dir Social Sciences Ontologies describe our categorization of the kinds of entities in the world—grouping them self Social Sciences 205 categorically by fundamental properties or characteristics.

Appendix D: Sample articles where we find definitions formatted as found in WoS.

Ref. Article Wheeler, TS. et al.. Feasibility and usability of an ontology-based mobile intervention for patients with hypertension. INTERNA- 004 TIONAL JOURNAL OF MEDICAL INFORMATICS, 119. NOV 2018. Jelokhani-Niaraki, M.. Knowledge sharing in Web-based collaborative multicriteria spatial decision analysis: An ontology-based 005 multi-agent approach. COMPUTERS ENVIRONMENT AND URBAN SYSTEMS, 72. NOV 2018. Chen, YQ. et al.. An ontology-based spatial data harmonisation for urban analytics. COMPUTERS ENVIRONMENT AND UR- 006 BAN SYSTEMS, 72. NOV 2018. Han, J. et al.. A computational tool for creative idea generation based on analogical reasoning and ontology. AI EDAM-ARTIFI- 007 CIAL INTELLIGENCE FOR ENGINEERING DESIGN ANALYSIS AND MANUFACTURING, 32(4). NOV 2018. Shiang, CW. et al.. Ontology reuse for multiagent system development through pattern classification. SOFTWARE-PRACTICE & 008 EXPERIENCE, 48(11). NOV 2018. Si, HY. et al.. Structured peer-to-peer-based publication and sharing of ontologies to automatically process SPARQL query on a 014 semantic sensor network. INTERNATIONAL JOURNAL OF DISTRIBUTED SENSOR NETWORKS, 14(10). OCT 8 2018. Larsen, RR. et al.. From Affective Science to Psychiatric Disorder: Ontology as a Semantic Bridge. FRONTIERS IN PSYCHIA- 015 TRY, 9(487). OCT 8 2018. Zhou, L. et al.. An ontology framework towards decentralized information management for eco-industrial parks. COMPUTERS & 016 CHEMICAL ENGINEERING, 118. OCT 4 2018. Alobaidi, M. et al.. Automated ontology generation framework powered by linked biomedical ontologies for disease-drug domain. 018 COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE, 165. OCT 2018. Hussain, M. et al.. Towards ontology-based multilingual URL filtering: a big data problem. JOURNAL OF SUPERCOMPU- 020 TING, 74(10). OCT 2018. Traverso, A. et al.. The radiation oncology ontology (ROO): Publishing linked data in radiation oncology using semantic web and 021 ontology techniques. MEDICAL PHYSICS, 45(10). OCT 2018. Gelbard, R. et al.. Sentiment analysis in organizational work: Towards an ontology of people analytics. EXPERT SYSTEMS, 35(5). 022 OCT 2018. Kang, Y. et al.. Disease Specific Ontology of Adverse Events: Ontology extension and adaptation for Chronic Kidney Disease. 023 COMPUTERS IN BIOLOGY AND MEDICINE, 101. OCT 1 2018. 025 Gao, W. et al.. Partial multi-dividing ontology learning algorithm. INFORMATION SCIENCES, 467. OCT 2018. Bharambe, U. et al.. Adaptive Pareto-based approach for geo-ontology matching. COMPUTERS & GEOSCIENCES, 119. OCT 027 2018. Knowl. Org. 47(2020)No.3 217 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

Ref. Article Biletskiy, Y. et al.. Building a business domain meta-ontology for information pre-processing. INFORMATION PROCESSING 032 LETTERS, 138. OCT 2018. Liang, JS.. An ontology-oriented knowledge methodology for process planning in additive layer manufacturing. ROBOTICS AND 033 COMPUTER-INTEGRATED MANUFACTURING, 53. OCT 2018. Alobaidi, M. et al.. Linked open data-based framework for automatic biomedical ontology generation. BMC BIOINFORMATICS, 036 19(319). SEP 10 2018. Rousseau, D. et al.. Systemic Semantics: A Systems Approach to Building Ontologies and Concept Maps. SYSTEMS, 6(3). SEP 037 2018. Takahashi, L. et al.. Redesigning the Materials and Catalysts Database Construction Process Using Ontologies. JOURNAL OF 044 CHEMICAL INFORMATION AND MODELING, 58(9). SEP 2018. Martin, J.. Ontology matters: a commentary on contribution to cultural historical activity. CULTURAL STUDIES OF SCIENCE 046 EDUCATION, 13(3). SEP 2018. Heimonen, J. et al.. Ontology Development for Patient Education Documents Using a Professional- and Patient-Oriented Delphi 047 Method. CIN-COMPUTERS INFORMATICS NURSING, 36(9). SEP 2018. Gibaud, B. et al.. Toward a standard ontology of surgical process models. INTERNATIONAL JOURNAL OF COMPUTER 050 ASSISTED RADIOLOGY AND SURGERY, 13(9). SEP 2018. Argentieri, MA.. Embodiment and Ontologies of Inequality in Medicine: Towards an Integrative Understanding of Disease and 060 Health Disparities. BODY & SOCIETY, 24(3). SEP 2018. Qiu, J. et al.. A hybrid-based method for Chinese domain lightweight ontology construction. INTERNATIONAL JOURNAL OF 061 MACHINE LEARNING AND CYBERNETICS, 9(9). SEP 2018. Zhu, XH. et al.. An Interoperable Model for the Intelligent Content Object Based on a Knowledge Ontology and the SCORM 063 Specification. JOURNAL OF EDUCATIONAL COMPUTING RESEARCH, 56(5). SEP 2018. Amith, M. et al.. Representing vaccine misinformation using ontologies. JOURNAL OF BIOMEDICAL SEMANTICS, 9(22). 066 AUG 31 2018. Tang, JL. et al.. Ontology Optimization Algorithm for Similarity Measuring and Ontology Mapping Using Adjoint Graph Frame- 068 work. ENGINEERING LETTERS, 26(3). AUG 28 2018. Kolyvakis, P. et al.. Biomedical ontology alignment: an approach based on representation learning. JOURNAL OF BIOMEDICAL 069 SEMANTICS, 9(21). AUG 15 2018. Zhong, BT. et al.. Ontology-based framework for building environmental monitoring and compliance checking under BIM envi- 070 ronment. BUILDING AND ENVIRONMENT, 141. AUG 15 2018. Shen, Y. et al.. EAPB: entropy-aware path-based metric for ontology quality. JOURNAL OF BIOMEDICAL SEMANTICS, 9(20). 072 AUG 10 2018. van Damme, P. et al.. From lexical regularities to axiomatic patterns for the quality assurance of biomedical terminologies and ontol- 075 ogies. JOURNAL OF BIOMEDICAL INFORMATICS, 84. AUG 2018. 076 Pozveh, ZH. et al.. FNLP-ONT: A feasible ontology for improving NLP tasks in Persian. EXPERT SYSTEMS, 35(4). AUG 2018. Orozova, D. et al.. Ontology Concept in Courses on Students. TEM JOURNAL-TECHNOLOGY EDUCATION MANAGE- 079 MENT INFORMATICS, 7(3). AUG 2018. Annane, A. et al.. Building an effective and efficient background knowledge resource to enhance ontology matching. JOURNAL 080 OF WEB SEMANTICS, 51. AUG 2018. Ansari, F. et al.. A problem-solving ontology for human-centered cyber physical production systems. CIRP JOURNAL OF MAN- 081 UFACTURING SCIENCE AND TECHNOLOGY, 22. AUG 2018. 085 Iliadis, A.. Algorithms, ontology, and social progress. GLOBAL MEDIA AND COMMUNICATION, 14(2). AUG 2018. Luttenberger, N. et al.. Standard International Trade Classification From Spreadsheet to OWL-2 Ontology. BUSINESS & INFOR- 089 MATION SYSTEMS ENGINEERING, 60(4). AUG 2018. Lin, R. et al.. Visualized Emotion Ontology: a model for representing visual cues of emotions. BMC MEDICAL INFORMATICS 096 AND DECISION MAKING, 18(64). JUL 23 2018. 218 Knowl. Org. 47(2020)No.3 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

Ref. Article Zhang, HS. et al.. An ontology-guided semantic data integration framework to support integrative data analysis of cancer survival. 097 BMC MEDICAL INFORMATICS AND DECISION MAKING, 18(41). JUL 23 2018. Roldan, ML. et al.. An Ontology-based Approach for Sharing, Integrating, and Retrieving Architectural Knowledge. ELEC- 102 TRONIC NOTES IN THEORETICAL COMPUTER SCIENCE, 339. JUL 11 2018. Freitas, V. et al.. Ontologia para representação de tempo no contexto de indicadores de desempenho. PERSPECTIVAS EM CIEN- 104 CIA DA INFORMACAO, 23(3). JUL-SEP 2018. Zheng, L. et al.. Complex overlapping concepts: An effective auditing methodology for families of similarly structured BioPortal 105 ontologies. JOURNAL OF BIOMEDICAL INFORMATICS, 83. JUL 2018. Louge, T. et al.. ASON: An OWL-S based ontology for astrophysical services. ASTRONOMY AND COMPUTING, 24. JUL 106 2018. Woznowski, PR. et al.. Activities of Daily Living Ontology for Ubiquitous Systems: Development and Evaluation. SENSORS, 107 18(7). JUL 2018. Moor, A de.. A Community Network Ontology for Participatory Collaboration Mapping: Towards Collective Impact. INFOR- 113 MATION, 9(7). JUL 2018. Dobreski, B. et al.. Ontology Informed Design to Advance Developers' Informal Online Learning. EDUCATIONAL TECHNOL- 118 OGY & SOCIETY, 21(3). JUL 2018. Sankari, SSU. et al.. Ontology-Enabled Generation of Simulation Software for a Complex Dynamic System. JOURNAL OF AER- 122 OSPACE INFORMATION SYSTEMS, 15(7). JUL 2018. Lima, R. et al.. OntoILPER: an ontology- and inductive logic programming-based system to extract entities and relations from text. 125 KNOWLEDGE AND INFORMATION SYSTEMS, 56(1). JUL 2018. Munnich, S.. Ontologien als semantische Zündstufe für die digitale Musikwissenschaft?. BIBLIOTHEK FORSCHUNG UND 126 PRAXIS, 42(2). JUL 2018. Roldan-Garcia, MD. et al.. Towards an ontology-driven clinical experience sharing ecosystem: Demonstration with liver cases. EX- 129 PERT SYSTEMS WITH APPLICATIONS, 101. JUL 1 2018. Akinyemi, A. et al.. An ontology-based data integration framework for construction information management. PROCEEDINGS 131 OF THE INSTITUTION OF CIVIL ENGINEERS-MANAGEMENT PROCUREMENT AND LAW, 171(3). JUN 2018. Eriksson, O. et al.. Institutional ontology for conceptual modeling. JOURNAL OF INFORMATION TECHNOLOGY, 33(2). 132 JUN 2018. Ruijgrok, P. et al.. ONTONAVSHOP: AN ONTOLOGY-BASED APPROACH FOR WEB-SHOP NAVIGATION. JOURNAL 136 OF WEB ENGINEERING, 17(3-4). JUN 2018. Ksiksi, A. et al.. Using Association Rules to Enrich Arabic Ontology. ENGINEERING TECHNOLOGY & APPLIED SCIENCE 137 RESEARCH, 8(3). JUN 2018. Arribas, B.. ONTOLOGY OF THE CURRENT NEWS: TECHNICAL AND NEW TECHNOLOGIES: BETWEEN CON- 139 TROL AND EMANCIPATION. ARIEL-REVISTA DE FILOSOFIA, (21). JUN 2018. 142 Lenzerini, M.. Managing Data Through the Lens of an Ontology. AI MAGAZINE, 39(2). SUM 2018. Gulic, M. et al.. Automatically Specifying a Parallel Composition of Matchers in Ontology Matching Process by Using Genetic Al- 143 gorithm. INFORMATION, 9(6). JUN 2018. Kramer, B.. Online music recommendation platforms as representations of ontologies of musical taste. COMMUNICATIONS- 148 EUROPEAN JOURNAL OF COMMUNICATION RESEARCH, 43(2). JUN 2018. Mucen, B.. THE ONTOLOGY OF CAPITAL: ON THE SHARED METHODOLOGICAL LIMITS OF MODERNIZA- 149 TION THEORY AND ITS CRITICS. HISTORY AND THEORY, 57(2). JUN 2018. Garcia-Weyandt, CM.. Mothers of Corn: Wixarika women, verbal performances, and ontology. ALTERNATIVE-AN INTERNA- 150 TIONAL JOURNAL OF INDIGENOUS PEOPLES, 14(2). JUN 2018. Engel, G. et al.. Ontology-Assisted Engineering of Cyber-Physical Production Systems in the Field of Process Technology. IEEE 151 TRANSACTIONS ON INDUSTRIAL INFORMATICS, 14(6). JUN 2018. Knowl. Org. 47(2020)No.3 219 L. M. Machado, M. B. Almeida, R. Rocha Souza. What Researchers are Currently Saying about Ontologies

Ref. Article Wang, CB. et al.. Ontology-driven data integration and visualization for exploring regional geologic time and paleontological infor- 152 mation. COMPUTERS & GEOSCIENCES, 115. JUN 2018. Parhi, M. et al.. A multi-agent-based framework for cloud service discovery and selection using ontology. SERVICE ORIENTED 154 COMPUTING AND APPLICATIONS, 12(2). JUN 2018. Hagedorn, TJ. et al.. A Knowledge-Based Method for Innovative Design for Additive Manufacturing Supported by Modular Ontol- 157 ogies. JOURNAL OF COMPUTING AND INFORMATION SCIENCE IN ENGINEERING, 18(2). JUN 2018. Ma, ZL. et al.. Ontology- and freeware-based platform for rapid development of BIM applications with reasoning support. AUTO- 161 MATION IN CONSTRUCTION, 90. JUN 2018. Gong, XF. et al.. A new method to measure the semantic similarity from query phenotypic abnormalities to diseases based on the 170 human phenotype ontology. BMC BIOINFORMATICS, 19(162). MAY 8 2018. Beniaminov, EM.. Ontology Libraries on the Web: Status and Prospects. AUTOMATIC DOCUMENTATION AND MATHE- 172 MATICAL LINGUISTICS, 52(3). MAY 2018. Salguero, AG. et al.. Ontology-based feature generation to improve accuracy of activity recognition in smart environments. COM- 175 PUTERS & ELECTRICAL ENGINEERING, 68. MAY 2018. McDaniel, M. et al.. Assessing the quality of domain ontologies: Metrics and an automated ranking system. DATA & 177 KNOWLEDGE ENGINEERING, 115. MAY 2018. Yago, H. et al.. ON-SMMILE: Ontology Network-based Student Model for MultIple Learning Environments. DATA & 178 KNOWLEDGE ENGINEERING, 115. MAY 2018. Maran, V. et al.. Domain content querying using ontology-based context-awareness in information systems. DATA & 179 KNOWLEDGE ENGINEERING, 115. MAY 2018. Mustafa, A. et al.. Integration of Heterogeneous Requirements using Ontologies. INTERNATIONAL JOURNAL OF AD- 180 VANCED COMPUTER SCIENCE AND APPLICATIONS, 9(5). MAY 2018. Guo, H. et al.. Ontology-Based Domain Analysis for Model Driven Pervasive Game Development. INFORMATION, 9(5). MAY 182 2018. Xu, FX. et al.. Developing an Ontology-Based Rollover Monitoring and Decision Support System for Engineering Vehicles. IN- 183 FORMATION, 9(5). MAY 2018. Schoenfisch, J. et al.. Root cause analysis in IT infrastructures using ontologies and abduction in Markov Logic Networks. INFOR- 193 MATION SYSTEMS, 74(SI). MAY 2018. Paredes-Valverde, MA. et al.. An ontology-based approach with which to assign human resources to software projects. SCIENCE 195 OF COMPUTER PROGRAMMING, 156. MAY 1 2018. Vitali, F. et al.. ONS: an ontology for a standardized description of interventions and observational studies in nutrition. GENES 197 AND NUTRITION, 13(12). APR 30 2018. Hoehn, JR. et al.. Students' flexible use of ontologies and the value of tentative reasoning: Examples of conceptual understanding in 205 three canonical topics of quantum mechanics. PHYSICAL REVIEW PHYSICS EDUCATION RESEARCH, 14(1). APR 11 2018. 206 Baker, N. et al.. Building a developmental toxicity ontology. BIRTH DEFECTS RESEARCH, 110(6). APR 3 2018. Sathiya, B. et al.. Automatic Ontology Learning from Multiple Knowledge Sources of Text. INTERNATIONAL JOURNAL OF 210 INTELLIGENT INFORMATION TECHNOLOGIES, 14(2). APR-JUN 2018. Merlo, G. et al.. Development and Validation of a Functional Behavioural Assessment Ontology to Support Behavioural Health 212 Interventions. JMIR MEDICAL INFORMATICS, 6(2). APR-JUN 2018. Coneglian, CS. et al.. Materialização da Web Semântica: um modelo de construção dinâmica de consultas baseados em mapeamento 213 de ontologias. PERSPECTIVAS EM CIENCIA DA INFORMACAO, 23(2). APR-JUN 2018. 214 Salguero, AG. et al.. Using Ontologies for the Online Recognition of Activities of Daily Living. SENSORS, 18(4). APR 2018.

220 Knowl. Org. 47(2020)No.3 K. Tharani. Just KOS!

Just KOS! Enriching Digital Collections with Hypertexts to Enhance Accessibility of Non-Western Knowledge Materials in Libraries†

Karim Tharani

University of Saskatchewan, 3 Campus Drive, Murray Building, Saskatoon, Saskatchewan, Canada S7N 5A4,

Karim Tharani has been an academic librarian at the University of Saskatchewan in Canada since 2009. He special- izes in digital information systems and also teaches at the University’s Edwards School of Business. Before joining academia, he was an information systems consultant at IBM Canada. His research interests include digital access and preservation, digital curation, database and linked data technologies, as well as the use of digital technologies to foster equity for marginalized knowledge. He is also Primary Investigator for Ginan Central, a major research project for long-term digital access and preservation of the ginans.

Tharani, Karim. 2020. “Just KOS! Enriching Digital Collections with Hypertexts to Enhance Accessibility of Non- Western Knowledge Materials in Libraries.” Knowledge Organization 47(3): 220-230. 21 references. DOI: 10.5771/0943-7444-2020-3-220.

Abstract: The knowledge organization systems (KOS) in use at libraries are social constructs that were conceived in the Euro-American context to organize and retrieve Western knowledge materials. As social constructs of the West, the effectiveness of library KOSs is limited when it comes to organization and retrieval of non-Western knowledge materials. How can librarians respond if asked to make non-Western knowledge materials as accessible as Western materials in their libraries? The accessibility of Western and non-Western knowledge materials in libraries need not be an either-or proposition. By way of a case study, a practical way forward is presented by which librarians can use their professional agency and existing digital technologies to exercise social justice. More specifically I demonstrate the design and development of a specialized KOS that enriches digital collections with hypertext features to enhance the accessibility of non-Western knowledge materials in libraries.

Received: 3 December 2019; Revised: 20 March 2020, 29 April 2020; Accepted: 28 April 2020

Keywords: ginans, material, community, digital, knowledge

† The author would like to thank Kristin Hoffmann, Associate Librarian at the University of Western Ontario, and the two anonymous reviewers for their very insightful suggestions and feedback.

1.0 Introduction KOSs tend to marginalize non-Western knowledge materials (Moulaison Sandy and Bossaller 2017). Moulaison While the phrase “knowledge is power” may be a truism, Sandy and Bossaller (2017) view this marginalization of having the skills to organize knowledge holds a similar non-Western materials in libraries as a social justice issue. agency. Librarians exercise their knowledge organization How can librarians respond to this issue of social justice (KO) skills for the public good by utilizing knowledge or- in libraries? One may answer this question by engaging in a ganization systems (KOSs) such as the Library of Congress debate, which might yield polarizing approaches from do- Classification (LCC), Library of Congress Subject Headings ing nothing at one end to replacing the underlying Western (LCSH), Dewey Decimal Classification (DDC), etc. While epistemology in library KOSs at the other end. The purpose many perceive these KOSs to be “universal systems,” library of this paper, however, is not to debate this question. In- Knowl. Org. 47(2020)No.3 221 K. Tharani. Just KOS! stead, I assert that the accessibility of Western and non-West- tioned before, the inability of libraries to provide equal foot- ern knowledge materials in libraries need not be an either-or ing to Indigenous knowledge materials, for instance, has proposition. I showcase a local solution in the form of a spe- been recognized as a social justice issue (Moulaison Sandy cialized KOS that respects local community norms and and Bossaller 2017). needs by taking advantage of existing library information Even with an unwavering conviction of librarians for so- technology infrastructure. By exploring a case study I fur- cial justice, the practicality of making library systems so- ther assert that librarians can and must exercise their profes- cially and cognitively just for marginalized knowledge mate- sional skills, courage and creativity to challenge systemic bi- rials remains daunting. Many librarians find themselves ask- ases in Western libraries in order to make library systems so- ing whether or not it is right for them to disrupt or disman- cially just for non-Western knowledge traditions and mate- tle existing library KOSs that work well for the majority of rials. library materials to accommodate marginalized materials. Librarians have been at the forefront of utilizing digital Finding a middle ground in the face of such a polarizing di- technologies as an equalizer in restoring social justice in li- lemma is perhaps the only practical way forward for librari- braries. With the rise of digital and web technologies in the ans, at least in the short term. It is, therefore, important for 1990s, most academic libraries, for instance, have been re- librarians to identify and appreciate the root causes of some sponsible for digitization and curation of unique and rare of the limitations of the existing library KOSs in charting a collections of their institutions. Digital collections have course forward. since become a de facto choice for organizing and sharing non-Western materials (Shiri 2006; Srinivasan et al. 2009). 2.1 Existing controlled vocabularies in library KOSs There are at least two reasons for the use of digital repositories for housing non-Western materials. First, among all the The fact that most KOSs in the West are primarily in Eng- systems in use at libraries, digital repositories remain the lish poses practical problems for non-Western communities most versatile in terms of ingesting materials of variable con- whose knowledge materials are seldom in English. Lan- tents and formats (text, sound, images, and multimedia). guage is essential to humans as social beings and is used to For cultures and communities that are not as dependent on communicate a shared worldview amongst communities. It literal practices for transmission of knowledge, digital repos- is by assigning words to concepts and objects that commu- itories are very accommodating for organizing their nities simplify and make sense of their complex reality. knowledge materials. Secondly, digital repositories impose Thus, vocabulary development, in any language, is a process minimal structure on the ingested materials in terms of KO of evaluating and organizing complex natural and artificial and allow users to organize their materials based on their phenomena with a particular point of view and purpose specific needs. While this flexibility has its advantages for (Izutsu 2009). It is not surprising then to understand why storing and categorizing materials, it also makes digital re- the use of LCSH, for example, is often criticized for its in- positories a poor tool for search and retrieval of materials. sensitivity toward non-Western worldviews. One of the This drawback, however, seems less applicable to non-West- ways librarians have addressed this issue in Indigenous com- ern materials as these materials mostly remain hidden and munities, for instance, is through initiatives like the Decol- inaccessible in libraries already. With this paper I show how onizing Description Project (Farnel et al. 2018). In this par- existing digital collections in libraries can be enriched with ticular initiative, members of the community were con- hypertext features to serve as specialized KOSs. sulted and empowered to develop local controlled vocabularies for inclusion in library KOSs. This approach is an ex- 2.0 Literature Review ample of how librarians have found a middle ground to accommodate complementary perspectives and priorities to Librarians are stewards of public knowledge. As responsible make tangible progress toward social justice for marginal- stewards, librarians must strive to make all knowledge avail- ized knowledge. able and accessible without prejudice and discrimination. However, library KOSs, particularly in the West, can be in- 2.2 Privileging commodified resources effective in the organization and retrieval of non-Western knowledge materials (Moulaison Sandy and Bossaller Over the years, libraries have worked with various stakehold- 2017). Some scholars argue that this ineffectiveness stems ers, including authors, publishers, vendors, aggregators and from the fact that all KOSs are social constructs, which lim- others, to ensure the efficiency of library systems in manag- its their effectiveness to a specific time, geography and coming Western materials. It is common to see Western munity of conception (Olsen 1998; Doyle et al. 2015; White knowledge resources packaged and indexed by vendors and 2018). The state of Indigenous knowledge in most libraries aggregators in ways that make it easier for libraries to dis- today is a stark exemplar of this ineffectiveness. As men- seminate knowledge materials through local catalogues and 222 Knowl. Org. 47(2020)No.3 K. Tharani. Just KOS! discovery systems. Nevertheless, this desire for efficiency Hypertextuality is “the ability to move without interrup- and the dependency on others have rendered libraries and tion from one information resource to another” (Feather librarians “complicit, if not responsible, for perpetuating and Sturges 2003, 232 quoted in Ridi 2018, 393). As this colonial approaches to knowledge by replacing traditional case study will demonstrate, the use of digital collections in knowledge with Western knowledge, especially in physical conjunction with hypertexts can provide the necessary digi- libraries established under colonial regimes” (Moulaison tal technology infrastructure to design and develop special- Sandy and Bossaller 2017, 132). Recognizing that it requires ized KOSs for non-Western knowledge materials. courage and perseverance to challenge established processes and biases in libraries, White (2018, 8) offers the following 3.0 Context: the community and the collection recommendations to librarians: For this case study, the author partnered with members of As librarians we know that our systems are imperfect the Ismaili community to understand the needs and aspira- and flawed, yet we often spend time waiting for other tions of the community to organize and safeguard its groups, like vendors or programmers to fix these is- revered tradition and collection of ginans (gnostic and de- sues instead of taking the initiative and doing it our- votional hymns). selves. The first step to creating truly international, non-colonial, and unbiased systems is to acknowledge 3.1 The Ismaili community that there are perspectives not currently represented in library KOSs. As a profession we need to stop pre- The Ismaili Muslim community, generally known as the Is- tending that geographical diversity always means cul- mailis, is a culturally pluralistic community that lives pri- tural diversity …. Finally, all cultures should be repre- marily in minority and diaspora settings around the world, sented equally within the systems we create and the including Canada, the United States, and the United King- voice of librarians from the cultures themselves dom. The majority of the community members settled in need[s] to be present from creation through revision. the Euro-American countries today trace their religious and cultural origins to the pre-colonized India. In India, the 2.3 Accepting the fallacy of “universal systems” community members were popularly known as the Khojas. The religious doctrine of the Khojas is known as Satpanth Library KOSs are generally perceived as “universal systems” (True Path), and their corpus of gnostic and devotional as they have been able to gather, describe, and organize ma- hymns is known as ginans. Beyond their liturgical role to- terials from various disciplines under increasingly unified day, ginans also remain deeply embedded in the personal and accessible interfaces. It is also evident that in the West- and social lives of the Satpanth Ismailis even today (Asani ern social context, library KOSs have been instrumental in 2002, 29): making disparate knowledge materials discoverable and accessible to patrons. Today, however, the social context and At a personal and family level, too, ginans are used in local needs are more diverse and expansive than in the past. many different contexts: individual verses can be Some argue that the notion of “universal systems” is unreal- quoted as proverbs; verses can be recited at homes to istic and must be replaced by local approaches that recog- bring baraka, spiritual and materials blessing; house- nize non-Western knowledge systems that remain marginal- wives, in a usage that stresses the links between the ized in libraries (Nakata et al. 2005). Thus, rather than ginans and folk tradition, often recite them while choosing to expand the definition of “universal systems” to working or as lullabies; audio cassettes can be found include non-Western knowledge materials, many librarians in many an Ismaili home and even cars! are advocating for designing specialized KOSs for libraries to serve local community needs by taking advantage of digi- 3.2 The collection of ginans tal technologies (Shiri 2006; Srinivasan et al. 2009; Chaudhry 2016; Farnel and Shiri 2018). Ginans were initially transmitted orally as early as the fifth One of the services that have helped librarians manage century, and the community members sang them in various non-Western knowledge materials is the digitization and Indian ragas (or melodies) to aid in memorization and digital collections service (Tharani 2019). This service, as transmission of ginans. Additionally, the Ismaili commu- Tharani notes (2019, 4), has proven useful for “librarians to nity also developed and utilized a particular script called collaborate with communities who are interested in sharing Khojki to transcribe ginans as early as the tenth century or safeguarding their history and heritage.” While a digital (Nanji 1978). Tracing its roots to mercantile communities repository is not a KOS in and of itself, it has the potential of India, Khojki is not a language in itself but a script that to be transformed into one by incorporating hypertexts. served as a shorthand to render ginans to text. The language Knowl. Org. 47(2020)No.3 223 K. Tharani. Just KOS! of the ginans itself is polyglottal (or multilingual) as it bor- 4.1 Gathering and analyzing materials rows vocabulary from Persian, Arabic, and several Indic languages (Shackle and Moir 1992). The community also This case study entailed gathering and analyzing 500 items founded the Khoja Press of India in 1903 to publish author- of ginans, which included manuscripts, lithographs, books, ized texts of ginans. The press used specialized German- CDs, as well as digital recordings. The items were primarily made fonts for the Khojki script for printing the ginan cor- shared by community members and digitized for online re- pus (Asani, 2011). Later on, as the community embraced trieval and access. Many unique Khojki items housed at the Gujarati as its lingua franca, a canon of ginans in Gujarati Harvard University Library were also included in the case also started to emerge in the mid-twentieth century. The ro- study. The contents of each selected item were further ana- manized transliterations of ginans became common during lyzed to identify and enumerate unique ginans. the 1970s as community members began to settle in the Euro-American countries. 4.2 Organizing knowledge materials With advances in technology, the Ismaili community members also started capturing sound recordings of the The materials gathered for the case study were divided into ginans on cassettes, CDs and digital audio files in the mid- five collections primarily based on their material type and to late- twentieth century (Asani 2002). The melodic aspect primary purpose, which are: “ginan archive,” “ginan canon,” of ginans is perhaps the primary reason why many of the “ginan commons,” “ginan recitals” and “ginan studies.” ginans continue to survive in the West. Kassam, an Ismaili These collections also acknowledge and continue the his- academic specializing in ginans, fondly laments the role of toric multifaceted efforts of the Ismaili community to safe- ginans during her childhood (2001, 1): “There seemed guard ginans for posterity. Materials in the “ginan archive” nothing unusual about singing ginans … on the way to the collection are primarily manuscripts in Khojki. These ar- beach, city, or school. As a child, I learned to articulate my chival materials serve as primary sources for the academic first requests to God and to express my first feelings of de- study of ginans. The “ginan canon” collection comprises votion and surrender through the language and music of materials that were officially published by the community ginans.” Thus, the corpus of ginans today include manu- to standardize the ginan corpus with changing times. Sev- scripts and lithographs in Khojki and Gujarati scripts, eral community elders and educators have also produced ro- printed transliterations of the ginans in Gujarati and Eng- manized English texts and translations of ginans which are lish alphabets, as well as analog and digital recordings of available in the “ginan commons” digital collection. The ginans. “ginan studies” collection includes a bibliography and texts of research studies on ginans. The audio recordings, curated 4.0 Methods in the “ginan recitals” collection, make up the majority of the items gathered in this study. A dedicated website was de- The multilingual and multimedia collection of ginans pre- veloped for each of the five independent digital collections, sented itself as a befitting candidate for this case study to de- as summarized in Table 1. velop a specialized KOS for non-Western knowledge materials. The case study has used the corpus of ginans to present 4.3 Identifying and ascribing metadata a pragmatic and hands-on account of how librarians can help enhance retrieval and access of non-Western materials Due to the diverse and dispersed nature of the items gath- by enriching digital collections with hypertext features. The ered, different types of information (metadata) attributes following is a summary of the process followed in this case about the materials were also captured for provenance, de- study. scription, retrieval, and access. While the metadata attrib-

Digital Collection Purpose Online Access Ginan Archive Collection of rare Khojki manuscripts of ginan texts. http://ginans.usask.ca/archive/ Ginan Canon Printed Khojki and Gujarati texts of ginans that are considered official. http://ginans.usask.ca/canon/ Ginan Commons Romanized texts of ginans in English developed by community individu- http://ginans.usask.ca/commons/ als and institutions. Ginan Recitals Audio recordings of recitations of ginans by community members across http://ginans.usask.ca/recitals/ the globe. Ginan Studies A comprehensive list and texts of academic research on ginans. http://ginans.usask.ca/studies/

Table 1. Digital collections of ginans. 224 Knowl. Org. 47(2020)No.3 K. Tharani. Just KOS! utes slightly varied amongst the five digital collections, the 5.1 The 5 Cs of specialized KOS—a conceptual design descriptive metadata for the ginans remained the same across all materials, as identified in Table 2. These metadata It may be useful to visualize a specialized KOS as a graph of attributes of ginans were primarily identified based on the nodes (units of information) and vertices (metadata attrib- practices prevalent in the Ismaili community. For instance, utes) that facilitates retrieval and access. In this visualiza- not all ginans have assigned titles, and it is a common prac- tion, the conceptual design of a specialized KOS consists of tice in the community to use the incipit (or the opening line a hierarchy of five decomposable nodes (or the 5 Cs), which of a ginan) as the title in such cases. Each ginan was assigned are: “community,” “collections,” “carriers,” “contents,” and a common identifier (master ID) for enumeration across all “containers.” The community node represents the highest the items, many of which had inconsistent wordings and level of organization to emphasize the local context of the transliteration conventions. KOS in serving specific community needs. The node directly below is the collections node, which represents the 5.0 Designing a specialized KOS for digital scope of materials organized within the specialized KOS. collections of ginans The next node in the hierarchy represents carriers or items (such as manuscripts, books, CDs) across all collections that Before proceeding with the account of designing the special- are included and served by the specialized KOS. The fourth ized KOS for ginans, it is vital to orient such a KOS in the node in the hierarchy represents contents and symbolizes existing typologies of KOSs. Scholars have suggested various the distinction between items and their contents (e.g., typologies of KOSs based on their purpose, function, and books and book chapters or albums and album tracks). The structure (Zeng 2008; Mazzocchi 2018). What can be seen in final node is represented by containers or digital collections, these typologies is the varying levels of semantic richness of which is where digital surrogates of materials are indexed KOSs, which can be as simple as lists of terms and as complex and stored for retrieval and access. In the case of ginans, this as ontologies. Thus, having the wherewithal to undertake se- node includes all the digital collections of ginans mentioned mantic analysis in a foreign language is a critical determinant before, including “Ginan Archive,” “Ginan Canon,” “Gi- of the effectiveness of a KOS. Given that librarians generally nan Commons,” “Ginan Recitals” and “Ginan Studies.” lack foreign language skills and resources to commission se- The relationships (or vertices) among these hierarchical mantic analysis, one can only expect specialized KOSs of non- nodes in the conceptual design of a specialized KOS are Western materials to be metadata-based lists and directories of linked together by various metadata attributes to facilitate attributes and authority files. The use of metadata in special- retrieval and access. The administrative metadata, for in- ized KOSs to replace semantic relations does not entirely ex- stance, is used for identifying and managing individual and onerate librarians of their responsibility toward non-Western institutional collections of the community. The descriptive materials. However, it does allow them to move forward in metadata pertain to information on various items that are enhancing the accessibility of marginalized knowledge mate- to be made available through the specialized KOS. The de- rials in their libraries. Using this simple approach to designing tails of the contents of an item are captured via the struc- specialized KOS also affords librarians and community mem- tural metadata, which can include details like title, order bers more flexibility to adjust the sophistication of specialized and location of intellectual works contained with a given KOSs from simple list structures to semantically rich ontolo- item. The reference metadata facilitate linking the intellec- gies as resources become available. tual knowledge organization of the materials with physical files. The information on digital collections and universal resource identifiers (URIs) assigned to various resources within the collections is part of the reference metadata suite.

Attributes Purpose master ID A unique numeric value assigned to each ginan for identification. incipit The starting line of untitled ginans. title (if any) Any additional name assigned to ginans by the composer or community. ginan type An attribute to distinguish between short and long titled ginans. composer Name of the composer of a ginan. state An attribute to distinguish between published and unpublished ginans. status An attribute to differentiate ginans with verifiable evidentiary sources to substantiate their existence and validity.

Table 2. Descriptive metadata attributes used for ginans. Knowl. Org. 47(2020)No.3 225 K. Tharani. Just KOS!

With the use of metadata in defining relationships between fined order or path for users to follow. The choice to access nodes in specialized KOSs, the notion of hypertextuality— linked information, and which order, is left to the users to the ability to decompose, describe, navigate, and retrieve in- decide. For example, multilinearity allows a user to access formation resources—is at the forefront of the conceptual English texts of a ginan in “Ginan Canon” and optionally design of specialized KOSs. The nodes and vertices have an navigate directly to the audio recitations of the same ginan overlay of hypertext features for navigation and retrieval of available in “Ginan Recitals.” This ability for a user to freely resources as depicted in Figure 1. discover, retrieve, and access resources through hypertexts According to Ridi (2018, 393), there are five fundamen- enriches these collections to function as a specialized KOS. tal characteristics of hypertextuality for KOSs: multilinear- Also, through the use of multilinearity, users can retrieve ity, granularity, integrability, interactivity, and multimedial- materials based on other facets (such as reciters, albums, and ity. These characteristics are also central to the design of the collections in the case of “Ginan Recitals” in Figure 2) with- specialized KOS developed for the ginans. out necessarily imposing any particular navigational path.

5.1.1 Multilinearity 5.1.2 Granularity

In hypertext, multilinearity refers to the ability to facilitate Granularity, according to Ridi (2018, 394), is a precondi- navigation of resources in any order, sequential or other- tion of hypertext and refers to the property of materials to wise. Although the number and nature of links embedded be “decomposed into smaller self-contained parts [while] are pre-determined, hypertexts do not impose any pre-de- still making sense and usable.” When working with Western

Figure 1. The 5 Cs of specialized KOS—a conceptual design. 226 Knowl. Org. 47(2020)No.3 K. Tharani. Just KOS!

Figure 2. The multilinear and contextual navigation across digital collections.

materials in libraries, for instance, librarians generally ability for users to navigate to particular items and collec- choose physical carriers such as books, serials, or CDs as still tions remain intact regardless of the extent of granularity. their preferred level of granularity for library KOSs. When working with non-Western knowledge, however, materials 5.1.3 Integrability may not come neatly packaged as their Western counterparts. Additionally, the community needs and norms may Ridi (2018, 394) asserts that integrability presents the user also influence the extent of granularity of decomposing and with the possibility of “following links in a hypertext by describing resources. In the case of the ginans, for example, moving from a node [or resource] to another … without the community members typically seek to retrieve individ- ever arriving to any definitive end.” A key enabler of inte- ual ginans regardless of their items (carriers). It was neces- grability for the ginan digital collections is the Ginan Mas- sary, therefore, to decompose the physical carriers of ginans ter Index (GMI). Consistent with the granularity set at in- by their contents. Figure 3 illustrates how a published item dividual ginans, the GMI provides an alphabetical registry in “Ginan Canon” has been decomposed based on its con- of ginans by enumerating and integrating all available mate- tents by using hypertexts. It is important to note that the rials across the five digital collections. Serving as the inter- Knowl. Org. 47(2020)No.3 227 K. Tharani. Just KOS!

Figure 3. Granularity in the specialized KOS of ginans. face of the specialized KOS for the ginans, the GMI, as ers, composers, and others (see Table 2 above) also provide shown in Figure 4, facilitates direct navigation and retrieval alternate interactive navigational paths. of specific resources by ginan titles. 5.1.5 Multimediality 5.1.4 Interactivity Multimediality refers to the versatility of materials that can An interactive specialized KOS provides users with multiple be retrieved and accessed by a given KOS. As mentioned be- discovery paths and puts them in control to freely choose fore, digital collections in libraries are versatile in terms of from the available choices. While standalone digital collec- ingesting multimedia resources. Most of the available open tions offer a certain level of interactivity to users in terms of and proprietary digital repository platforms have built-in accessing textual and multimedia resources, they generally support for standard media and file types, including PDF have limited and fixed paths of interaction. For instance, to and MP3 formats. In the case of the ginans, scans of all dig- access a specific ginan from a particular item in a digital col- itized textual materials were converted to PDF files. Because lection, a user would typically need to page through several community members recite ginans in various tunes, it is images to arrive at the desired ginan. With a hypertext-ena- possible to digitally capture and store these melodies as MP3 bled GMI interface, users can traverse through multiple files. Table 3 provides a summary of files by type accessible paths in discovering collections, items, as well as ginans across the five ginan digital collections. housed across multiple digital collections. Additionally, the descriptive metadata attributes such as contributors, recit- 228 Knowl. Org. 47(2020)No.3 K. Tharani. Just KOS!

Figure 4. Ginan Master Index (GMI).

Digital Collection Media Type Number of Resources Ginan Archive Portable Document Format (PDF) 914 Ginan Canon Portable Document Format (PDF) 1243 Ginan Commons Portable Document Format (PDF) 2610 Ginan Recitals MP3 audio 5104 Ginan Studies Portable Document Format (PDF) 250

Table 3. Summary of formats and files by digital collections.

6.0 Discussion and analysis 6.1 Accommodating foreign scripts and languages

Any KOS is a social construct that derives its value and util- One of the challenges that librarians must surmount when ity based on the geography, culture, time, and community it working with non-Western knowledge resources is the lan- is conceived in and for. Many of the KOSs in use in libraries guage. For these materials to be available through existing were conceived in the Euro-American context to serve the library KOSs, access to standard romanization guides (for organization and retrieval needs of the West (White 2018). transliterating foreign language using English alphabets) is With this case study I have presented a practical way to ad- crucial. These guides can then be used in specialized KOS to dress this challenge, without necessarily disrupting existing consistently capture descriptive metadata about the library processes that continue to work well for Western knowledge materials using an agreed-upon convention. The knowledge materials. I showcased how librarians can still case of the Ismaili community was particularly complex as enhance access to non-Western materials by devising practi- the script used by the community to write ginans was dis- cal KOSs that utilize existing library information technol- tinct from the language of the ginans. In general, however, ogy infrastructure. Some of the general insights and lessons non-Western materials are often in a foreign language that gleaned from this case study are summarized below. may already have a standardized convention for romaniza- Knowl. Org. 47(2020)No.3 229 K. Tharani. Just KOS! tion. The engagement of community members to lend their footing to non-Western knowledge materials. After a very language expertise can also be extremely helpful for address- brief discussion on the merits of creating local KOSs for dig- ing this challenge. ital collections of non-Western knowledge materials, a conceptual design identifying the 5 Cs of specialized KOSs was 6.2 Embracing plurality of sources and resources discussed using Ridi’s characteristics of hypertextuality as a framework. The implementation of the conceptual design As mentioned before, non-Western knowledge materials was also presented for the tradition of ginans (a collection typically do not come packaged with well-defined metadata of devotional and gnostic hymns) of the Ismaili community attributes. Consequently, the contents in available items can as a case study. This specialized KOS for the ginans was be repeated and reproduced within a community in multi- based on the division of a corpus of 500 source materials ple forms and media. It can require extensive effort on the into five distinct digital collections, which were enriched part of librarians to identify and de-duplicate unique con- with hypertexts. tents (works). Consultation and collaboration in this regard can not only lead to better results but also can save time and References resources. An additional benefit of engaging the community is that the elders and educators in the community can Asani, Ali S. 2002. Ecstasy and Enlightenment: The Ismaili share their domain knowledge in terms of gauging sensitiv- Devotional Literature of South Asia. London: I. B. Tauriss. ities surrounding materials that can and cannot be shared Asani, Ali S. 2011. “From Satpanthi to Ismaili Muslim: The publicly. Librarians can reciprocate these community ef- Articulation of Ismaili Khoja Identity in South Asia. A forts by lending their professional expertise in organizing Modern History of the Ismailis.” In A Modern History of the materials for enhanced accessibility and preservation. the Ismailis: Continuity and Change in a Muslim Com- munity, ed. Farhad Daftary. London: I. B. Tauris Pub- 6.3 Describing works in addition to items lishers, 95-128. Chaudhry, Abdus Sattar. 2016. “Re-conceptualization of One of the insights gained from this project was the necessity Knowledge Organization: Imperatives of Networked Re- to disambiguate carriers (items) from the contents (works) in sources and Digitization.” International Journal of the collections. Although the underlying library standards Knowledge Content Development & Technology 6: 93-107. and systems are capable of cataloguing at multiple levels of Doyle, Ann M., Kimberley Lawson and Sarah Dupont. 2015. granularity, cataloguing items of Western knowledge materi- “Indigenization of Knowledge Organization at the als has become a best practice in many libraries. When it Xwi7xwa Library.” Journal of Library and Information comes to organizing non-Western knowledge, the prevailing Studies 13: 107-34. https://circle.ubc.ca/handle/2429/ best practice of cataloguing carriers may need to be adjusted 54261 based on the local needs and norms of the community. Farnel, Sharon and Ali Shiri. 2018. “Community-Driven Knowledge Organization for Cultural Heritage Digital Li- 6.4 Incorporating local community needs braries: The Case of the Inuvialuit Settlement Region.” Advances in Classification Research Online 29: 9-12. The plurality of materials gathered for this case study rein- Farnel, Sharon, Denise Koufogiannakis, Ian Bigelow, Anne forced the diverse needs and audiences for ginans. For any Carr-Wiggin, Debbie Feisst, Kayla Lar-Son and Sheila specialized KOS to be acceptable and useful for the commu- Laroque. 2018. “Unsettling Our Practices: Decolonizing nity, librarians must commit to putting the needs of the Description at the University of Alberta Libraries.” In- community at the forefront. This commitment requires ternational Journal of Information, Diversity, & Inclu- building online interfaces with multiple views and access sion 2, no. 1/2. https://jps.library.utoronto.ca/index. points for retrieval and access of materials. php/ijidi/article/view/32218/24644/ Feather, John and Paul Sturges. 2003. International Ency- 7.0 Conclusion clopedia of Information and Library Science, ed. John Feather and Paul Sturges. 2nd ed. London: Routledge. Librarians have equal responsibility toward the accessibility Izutsu, Toshihiko. 2002. Ethico-Religious Concepts in the of dominant and marginalized materials in their libraries. Qur'an. Montreal: McGill University Press. However, librarians need not think of these two as compet- Kassam, Tazim R. 2001. “Teaching Religion and Music: ing choices but as necessary complements. By way of a case Those Who Sing Pray Twice.” Spotlight on Teaching 16, study, a practical way forward was presented to enable li- no. 2: 1-12. brarians to use their professional agency and existing library Mazzocchi, Fulvio. 2018. “Knowledge Organization System information technology infrastructure to provide equal (KOS): An Introductory Critical Account.” Knowledge 230 Knowl. Org. 47(2020)No.3 K. Tharani. Just KOS!

Organization 45: 54-78. doi:10.5771/0943-7444-2018- Shiri, Ali. 2006. “Knowledge Organization Systems in Cana- 1-54 dian Digital Library Collections.” Proceedings of the An- Moulaison Sandy, H. and Bossaller, J. 2017. “Providing Cog- nual Conference of CAIS/Actes du congrès annuel de l’ACSI. nitively Just Subject Access to Indigenous Knowledge doi:10.29173/cais189 through Knowledge Organization Systems.” Cataloging Srinivasan, Ramesh, Robin Boast, Jonathan Furner and & Classification Quarterly 55, no. 3: 129-52. doi:10.1080/ Katherine M. Becvar. 2009. “Digital Museums and Di- 01639374.2017.1281858 verse Cultural Knowledges: Moving Past the Traditional Nakata Martin, Alex Byrne, Vicky Nakata V. and Gabrielle Catalog.” The Information Society 25: 265-78. Gardiner. 2005. “Indigenous Knowledge, the Library Tharani, Karim. 2019. “Shifting Established Mindsets and and Information Service Sector, and Protocols.” Austral- Praxis in Libraries: Five Insights for Making Non-West- ian Academic & Research Libraries 36, no. 2: 7-21. ern Knowledge Digitally Accessible Through Commu- Nanji, Azim. 1978. The Nizārī Ismāʻīlī Tradition in the Indo- nity Engagement.” Canadian Journal of Academic Li- Pakistan Subcontinent. Delmar, NY: Caravan Books. braries 4: 1-13. doi:10.33137/cjal-rcbu.v4.30018 Olson, H. A. 1998. Mapping Beyond Dewey's Boundaries: White, Hollie. 2018. “Decolonizing the Way Libraries Organ- Constructing Classificatory Space for Marginalized ize.” Paper presented at the IFLA WLIC 2018. http:// Knowledge Domains. Library Trends 42, no. 2: 233-54. library.ifla.org/2221/1/207-white-en.pdf Ridi, Riccardo. 2018. “Hypertext.” Knowledge Organiza- Zeng, Marcia Lei. 2008. “Knowledge Organization Systems tion 45: 393-424. (KOS).” Knowledge Organization 35: 160-82. Shackle, Christopher and Zawahir Moir. 1992. Ismaili Hymns from South Asia: An Introduction to the Ginans. London: SOAS.

Knowl. Org. 47(2020)No.3 231 R. Szostak. Basic Concepts Classification (BCC)

The Basic Concepts Classification (BCC)†

Rick Szostak

University of Alberta, 8-19 Tory Building, Edmonton, AB, T6G 2H4, Canada,

Rick Szostak is Professor and Chair of Economics at the University of Alberta. He was elected President of the International Society for Knowledge Organization (ISKO) in July 2018. He is the author of fifteen books and fifty journal articles in economics, history, interdisciplinary studies, information science, and several other fields. He has studied the theory and practice of interdisciplinarity for two decades and has emphasized in the last decade the ways in which knowledge organization systems might better facilitate interdisciplinary research and teaching. He co- authored Interdisciplinary Knowledge Organization in 2016 and has created the Basic Concepts Classification (BCC).

Szostak, Rick. 2019. “Basic Concepts Classification (BCC).” Knowledge Organization 46(3): 231-243. 38 references. DOI:10.5771/0943-7444-2020-3-231.

Abstract: The Basics Concept Classification (BCC) is a “universal” scheme: it attempts to encompass all areas of human understanding. Whereas most universal schemes are organized around scholarly disciplines, the BCC is instead organized around phenomena (things), the relationships that exist among phenomena, and the properties that phenomena and relators may possess. This structure allows the BCC to apply facet analysis without requiring the use of “facet indicators.” The main motivation for the BCC was a recognition that existing classifications that are organized around disciplines serve interdisciplinary scholarship poorly. Complex concepts that might be understood quite differently across groups and individuals can generally be broken into basic concepts for which there is enough shared understanding for the purposes of classification. Documents, ideas, and objects are classified synthetically by combining entries from the schedules of phenomena, relators, and properties. The inclusion of separate schedules of—generally verb-like—relators is one of the most unusual aspects of the BCC. This (and the schedules of properties that serve as adjectives or adverbs) allows the production of sentence-like subject strings. Documents can then be classified in terms of the main arguments made in the document. BCC provides very precise descriptors of documents by combining phenomena, relators, and properties synthetically. The terminology employed in the BCC reduces terminological ambiguity. The BCC is still being developed and it needs to be fleshed out in certain respects. Yet it also needs to be applied; only in application can the feasibility and desirability of the classification be adequately assessed.

Received: 19 July 2019; Revised: 1 August 2019; Accepted 6 August 2019

Keywords: Basic Concepts Classification, BCC, classification, subject relators, phenomena, concepts

† Derived from the article of similar title in the ISKO Encyclopedia of Knowledge Organization, Version 1.0 published 2019-07-17. Article category: KOS, specific (general/universal).

1.0 Introduction sess. As we shall see below, this structure allows the BCC to apply facet analysis—an approach to classification long em- The Basics Concept Classification (BCC) is a “universal” phasized within the field of knowledge organization—with- scheme: it attempts to encompass all areas of human under- out requiring the use of “facet indicators”1 to tell the user standing. Diverse users—both general users from different which facet is being addressed by a particular notation. cultural backgrounds and scholars from across disciplines— can potentially utilize the BCC to find documents, objects, 1.1 Motivation or ideas produced in any culture or discipline. Whereas most universal schemes are organized around scholarly dis- The main motivation for the BCC was a recognition that ciplines, the BCC is instead organized around phenomena existing classifications that are organized around disciplines (things), the relationships that exist among phenomena, serve interdisciplinary scholarship poorly. The fact that dif- and the properties that phenomena and relators may pos- ferent terminology and organizing principles are pursued 232 Knowl. Org. 47(2020)No.3 R. Szostak. Basic Concepts Classification (BCC) for different disciplines hampers the research and commu- of.” Several more categories have been added to address nat- nications of interdisciplinary scholars and students. Over ural science subject matter: in accord with the idea of inte- time, a variety of other advantages of a classification grative levels these address “waves and particles,” “molecules grounded in phenomena, relationships and properties have and atoms,” “rocks,” “biological entities,” “flora and been appreciated. Such a classification can achieve greater fauna,” “celestial objects,” and “mathematical concepts.” precision even within disciplines than existing classifica- Schedules of relators were added over time, based on Szostak tions achieve and can be used across galleries, libraries, ar- (2012), and these were given quite different notation from chives and museums (the so-called GLAM sector) and in- the schedules of phenomena. A schedule of “properties” deed across any organized repository of information. (We was developed inductively over time as the BCC was used to shall see in Section 2.4 below that a classifier employing classify sets of documents and objects; it also is notationally BCC moves fairly directly from a sentence in an object or distinct. The notations employed for phenomena, relators, document description to a BCC subject string and can, and properties are described in sections 2.1, 2.2, and 2.3 re- therefore, treat any document or object or idea that can be spectively below. We can recognize here that one key goal in described in a sentence.) This is especially important as users assigning notation was to ensure that these three quite dif- increasingly seek information across a variety of different ferent types of schedule were easily distinguished. Other databases and, at present, struggle to master different classi- goals were to have very short notations so that multiple fications and search interfaces for each. Such a classification terms could be combined in subject strings. Wherever pos- is useful not just for documents but for objects (thus its util- sible, notation was developed that would be easy for users ity for museums and galleries—Szostak 2016, 2017) and and classifiers to remember. ideas (and thus its utility for a variety of databases). It may As Szostak developed the scheme, he was encouraged prove advantageous for the semantic web, since its separate and informed by preceding efforts to develop a universal schedules of things, properties, and relators are well-suited phenomenon-based classification within the field of to the (subject)(predicate or property)(object) structure knowledge organization (Szostak, Gnoli, and Lopez-Huer- employed on the semantic web. (Szostak, Gnoli, and Lopez- tas 2016, 96-100, review this history). The Classification Huertas 2016, 177-82). Research Group in the United Kingdom had discussed the principles of such a scheme in much detail over a period of 1.2 History of the BCC many years (e.g., Classification Research Group 1969). The most important influence, though, has been Claudio Gnoli, Szostak (2004) speculated on the value for interdiscipli- who has pioneered the Integrative Levels Classification narity of a classification system grounded in phenomena ra- (ILC; 2004-; Gnoli 2016; 2017a; 2017b; 2018). There are ther than disciplines. That book also reprised a three-page many similarities between the ILC and BCC; both take a hierarchical table of the main subjects addressed by human faceted phenomenon-based and analytic-synthetic ap- scientists. Szostak (2004) was aimed primarily at practicing proach to classification. Szostak has contributed to the ILC interdisciplinary scholars. Over the next years, Szostak pub- project, learned much from it, and co-authored several pa- lished a series of articles in the Journal of Documentation, pers and a book (the latter with and inspired by Maria Journal of the Association for Information Science and Tech- Lopez-Huertas) with Claudio Gnoli. The major initial dif- nology (JASIST), Library Trends, and especially Knowledge ference between the BCC and ILC was the development of Organization, which sought to both justify and describe the separate schedules of relators and properties; this has led approach to classification taken in the BCC. At the same over time to an emphasis within BCC on the development time, the BCC was fleshed out to reflect the principles out- of subject strings that follow grammatical rules (see below). lined in these various publications. The BCC website2 dis- Gnoli, Szostak, Lopez-Huertas, and others also collaborated cusses how each of about two dozen articles and books sup- on the León Manifesto, issued after the ISKO-Spain confer- ports the project of BCC. ence in León in 2007 to urge an approach to classification The classification of phenomena expanded on the table more sympathetic to interdisciplinary scholarship. Szostak developed in Szostak (2004), which itself had reflected the also benefitted from theoretical discussions with Birger reading of hundreds of works across all human science dis- Hjørland and others over the years (see Fox 2012 for a disciplines in Szostak (2003). Ten main categories of phenom- cussion). Szostak and Richard Smiraglia have collaborated ena were identified: 1) two individual-level categories of in recent years on a series of papers and a research grant that “genetic predisposition and individual differences” and (among other things) compare the BCC to both the Uni- “non-human environment;” and, 2) seven societal-level cat- versal Decimal Classification and the Linked Open Data egories, “art,” “culture,” “economy,” “health and popula- cloud: This research has inspired much further clarification tion,” “politics,” “social structure,” and “technology and and testing of the BCC (Smiraglia and Szostak 2018, Szos- science.” These were subdivided logically in terms of “type tak et al. 2018). Knowl. Org. 47(2020)No.3 233 R. Szostak. Basic Concepts Classification (BCC)

Documents, ideas, and objects are classified synthetically Philosophers have long debated the nature of concepts by combining entries from the schedules of phenomena, re- and whether it is possible for diverse individuals to have lators, and properties. It was recognized eventually that the shared understandings of concepts. Szostak (2011) reviewed resulting subject strings resembled sentences or sentence leading concept theories and argued that most if not all of fragments. Szostak, in a series of publications (especially these supported the idea of basic concepts. He stressed that 2017), then explored the advantages of pursuing a grammat- philosophers often focused on whether precise definitions of ical approach to classification. The BCC and the rules em- concepts were possible; classificationists can be satisfied with ployed for synthesizing subject strings were adapted to pur- a more relaxed standard: can enough similarity in understand- sue purposely a common grammatical form. One key ad- ing of a concept be achieved such that users of the classifica- vantage of doing so is that a user query entered grammati- tion can be guided to relevant documents, objects, or ideas? cally can be readily translated into an appropriate and rele- Classificationists debate whether to employ precise jar- vant subject string. gon in classifications—at the cost of users being unfamiliar The BCC has been developed online at: https://sites. with terminology—or instead employ natural language— google.com/a/ualberta.ca/rick-szostak/research/basic-con- with the danger of ambiguity. Reliance on basic concepts al- cepts-classification-web-version-2013/guiding-principles. lows us to have the best of both worlds, for there is consid- Many additions have been made to the BCC online since it erable shared understanding of the meaning of basic con- was uploaded in 2013. The schedules for the subject matter cept terminology as employed in everyday language. of the human sciences are largely complete (with some exceptions such as the treatment of mental disorders), but 1.4 Guiding principles work remains to be done on several schedules addressing natural science phenomena. The schedules of relators and We have already mentioned some key guiding principles: properties are well developed. The classification has been successfully employed in classifying dozens of randomly se- – Synthetic classification utilizing separate schedules of lected documents and museum artifacts. The BCC is being phenomena, relators, and properties translated into linked open data. Other ongoing develop- – Short logical hierarchies of phenomena ments are discussed toward the end of this article. – Reliance wherever possible on “basic concepts:” those that are understood in broadly similar ways across indi- 1.3 The nature of basic concepts viduals and communities – Synthetic subject strings generally resemble sentences or Scholars of knowledge organization worry about whether sentence fragments. The use of common grammatical different individuals and groups can have the same under- format facilitates search; a user’s search query can be standing of terminology. One advantage of a discipline- translated into the most relevant subject string. grounded approach to classification is that scholars within a discipline may have similar understandings of the termi- Some other key guiding principles can be mentioned: nology employed in their discipline. Szostak (2011) argued that complex terms that might be understood differently by – For relators, several dozen key relators are developed different groups or individuals could be broken into basic within a handful of very flat hierarchies. These can be concepts for which there was a broadly shared understand- combined synthetically with each other and with phe- ing across groups and individuals. “Globalization” might be nomena or properties to generate thousands of very pre- ambiguous but “exports” (of goods) was likely to be under- cise relators. stood in a broadly similar way by most individuals. This – Hierarchies of phenomena generally reflect a “type of” shared understanding need not be perfect in order for quite disaggregation (rarely “elements of”), which reflects an different individuals to be guided to relevant documents (or ontological understanding of the world, supplemented objects or ideas) by the same classification system. Basic as necessary by literary warrant, ensuring that all relevant concepts are defined then as concepts for which there is concepts are captured. enough of a shared understanding for the purposes of clas- – The notations attached to concepts are generally both sification. The BCC has been constructed wherever possi- short and expressive. It is thus possible to synthesize several ble in terms of basic concepts. It is hoped in future to test terms and still have a notation for a subject string of man- the hypothesis that diverse users have broadly similar under- ageable length. A user familiar with the classification may standings of BCC terminology. Scope notes could be pro- be able to recognize the string from the notation. vided to clarify terminology in cases (such as political ideo- – Detailed classifications of methods and theory types are logies) where it is difficult to reduce ambiguity to the de- included in the classification of things so that works can sired level. be precisely classified in terms of the theories and meth- 234 Knowl. Org. 47(2020)No.3 R. Szostak. Basic Concepts Classification (BCC)

ods applied as well. Scholars often care not just (or pri- 1.6 The BCC and the tradition of facet analysis marily) about what a work is about but what theories and methods were applied. It is, at present, generally im- For decades, the knowledge organization literature has advo- possible to search by theory or method applied. cated facet analysis (see Hjørland 2013). That is, classification systems should seek to capture the various facets by 1.5 Advantages of separate schedules of relators which documents might be distinguished. Facet analysis encourages a synthetic approach to classification in which dif- The inclusion of separate schedules of—generally verb- ferent facets are combined in a subject string. The BCC cap- like—relators is one of the most unusual aspects of the tures each of the facets commonly identified in the literature; BCC. This (and the schedules of properties that serve as ad- these turn out to be either particular types of phenomena jectives or adverbs) allows the production of sentence-like (easily identified in the BCC schedules) or particular ele- subject strings. Documents can then be classified in terms ments of a grammatically constructed subject string. In a sen- of the main arguments made in the document. The classi- tence fragment such as (dogs)(bite)(mail carriers), it is clear fier can move directly from a sentence in an abstract or book that “dogs” is the subject, “bite” is the operation, and “mail description to a subject string (Rules have been developed carrier” is the object, as these facets are defined by the Bliss’s in Szostak 2017 to encourage the development of subject Bibliographic Classification3 (BC2) (Bliss n.d.). Szostak strings of typical grammatical format; see below). Since (al- (2017) describes how each facet identified by Bliss is identi- most) all humans spend their lives thinking and talking in fied grammatically. The classifier need not explicitly per- sentences, they can more readily comprehend subject form facet analysis but merely translates a sentence(s) from a strings that follow a grammatical format. Moreover, a search document description into a grammatical subject string. interface can potentially move easily from a search query in Sentences, we might note, are also faceted and synthetic, but the form of a sentence to a relevant subject string that re- we do not need facet indicators to communicate sentences; sembles a sentence or sentence fragment. we immediately recognize the purpose of a verb without hav- The BCC can be used to classify ideas as well as docu- ing to be told that it is a verb. By mimicking sentence struc- ments. Ideas are generally expressed in the form of sen- ture, the BCC pursues facet analysis without facet indica- tences. They are hard to capture accurately unless a classifier tors. The fact that the BCC schedules of things is organized has recourse to schedules of relators (and properties). Less around phenomena rather than disciplines is also important obviously, the BCC is also well-suited to the classification for it allows us to indicate certain facets such as “material” of objects. Again, a classifier can often move fairly quickly (again see Bliss n.d.) by simply employing the notation for from a key sentence in an object description toward a sub- the single schedule of materials within BCC. ject string (for example (axe)(for)(war)), and user queries Other classifications that take a faceted approach, such formulated grammatically can be translated into the rele- as the Integrative Levels Classification (2004), employ vant subject string. “facet indicators:” notations that inform the user which The existence of schedules of relators facilitates the logi- facet a particular piece of notation refers to. A subject string cal classification of phenomena. It is inconvenient, for ex- may then devote several notational spaces to facet indica- ample, to treat “recycling” in a classification that only ad- tors. Within a BCC subject string, facets are indicated either dresses phenomena. Recycling is something that is done to by a particular schedule of phenomena (such as materials) phenomena rather than a type of phenomenon. One cannot or a particular grammatical role (such as a verb). BCC sub- treat a process such as “recycling” as a logical “type of” subject strings can thus save notational space by eschewing facet class of a thing (Mazzocchi et al. 2007). In the BCC, recy- indicators. cling is treated as a relator rather than a phenomenon. See Julien et al. (2013) for a discussion of hierarchical complex- 1.7 Philosophical justification of the BCC ity in existing classification systems. It is often maintained in the knowledge organization lit- Szostak, Gnoli, and Lopez-Huertas (2016) summarizes an erature that different people might develop different classi- extensive literature that argues for both the feasibility and fications of phenomena. A classic example is that chemists desirability of a phenomenon-based comprehensive (“uni- may prefer a different classification of chemical elements to versal”) classification. Philosophical justifications of the that preferred by pharmacologists. In the BCC, chemicals BCC in particular can be found in Szostak (2008; 2011; are classified as chemists would classify them. The interests 2012; 2014a; 2014b; 2015a; and 2017). of pharmacologists are captured synthetically; for example, Szostak (2011) made the key argument that complex (chemical)(for reducing)(blood pressure). This is only pos- concepts that might be understood quite differently across sible because of the schedules of relators that can connect groups and individuals could generally be broken into basic chemicals and physiological outcomes. concepts for which there was enough shared understanding Knowl. Org. 47(2020)No.3 235 R. Szostak. Basic Concepts Classification (BCC) for the purposes of classification. Szostak argued that this urges classification within communities of shared under- hypothesis was consistent with most if not all of the major standing—could be interpreted and applied in a manner that philosophical theories of concepts. He argued that practi- facilitated cross-disciplinary classification. Szostak (2010) ar- tioners of knowledge organization should seek consistency gued that domain analysis could be applied in comprehen- with a broad range of philosophical theories. He noted that sive classification. That paper also applied work process anal- for many philosophers, basic concepts represented real ysis to show the advantages for interdisciplinary scholarship things (or relationships) that people observe (similarly) in of a classification such as BCC. Szostak (2012) developed the world around them. He urged a pragmatic and empiri- and justified the classification of relators employed in the cal approach; knowledge organization need not seek an un- BCC. A key argument here is that most works discuss rela- attainable elimination of terminological ambiguity but tionships among two or more phenomena (things) and are must merely explore whether ambiguity can be reduced to thus best classified synthetically using relators. Szostak the level needed for particular classificatory projects. (2014a) argued that an approach such as that taken in the Szostak (2015a) argued that any classification should be BCC could address many of the challenges noted in the lit- judged in terms of a variety of both philosophical and prac- erature around social diversity. The BCC might thus facili- tical criteria. It argued that (a classification like) the BCC tate cross-group respect and understanding just as it is de- best reflects the nature of the works classified, since it cap- signed to facilitate interdisciplinary communication. Szostak tures the key arguments made in a work. Smiraglia (2001) (2014b) discussed how a classification such as BCC can im- had argued that the “nature of a work” should primarily be prove the scholarly enterprise. In particular, by using a com- understood as the key ideas that the work contains; Szostak mon vocabulary to classify works in terms of their key argu- built upon this insight to stress the importance of sentences ments, the BCC can identify precise conflicts between au- that would generally contain causal arguments. The BCC thors or disciplines (but also apparent conflicts that reflect also signals to users the likely importance of a work (by po- terminological confusion rather than real differences of tentially classifying theory, method and perspective applied, opinion) and often the sources of these. Knowledge organi- and the precise causal arguments investigated), addresses zation could then serve as an antidote to extreme skepticism several challenges associated with hierarchy, meets ethical about the possibilities of scholarly understanding; identify- standards, responds to various concerns raised by the lived ing conflicts and recognizing their sources are the first steps experience of librarians, and is able to attach formal defini- toward transcending those conflicts. Szostak (2017 and else- tions to natural language concepts. where) argued that the BCC addressed a range of theoretical The knowledge organization literature has often con- and practical concerns in the classification of museum arti- trasted pre-coordinated classifications in which the classi- facts. That article also explored the advantages of utilizing fier must choose from a set of (usually complex) subject subject strings that follow standard grammatical formats. terms with post-coordinated classifications in which the classifier can synthesize terms to create novel subject head- 2.0 The structure of the Basic Concepts ings. Though a variety of advantages have been attributed Classification to the post-coordinated approach taken by classifications such as the BCC, it has generally been argued that pre-coor- The BCC has separate schedules of phenomena, relators, dinated systems allow greater precision. It is feared that a and properties. post-coordinated search for (history)(of)(philosophy) will yield many (irrelevant to the search) works on philosophy of 2.1 The classification of phenomena history (Sauperl 2009). Szostak (2015a) argued that this is only true if we insist on search interfaces that do not dis- Though the vast majority of general (i.e., universal) classifi- criminate based on the word order in search queries. Szostak cation systems employed in the world are organized around had a group of computer science students develop a search disciplines, the idea of organizing instead around phenom- algorithm that did prioritize the order of terms in the search ena (i.e., things) has long been advocated within the field of query. A search for (history)(of)(philosophy) does then knowledge organization. The history of the idea of organiz- yield works on history of philosophy rather than philosophy ing by phenomena is reviewed in Szostak, Gnoli and Lopez- of history. With the right search interface, then, the BCC Huertas (2016, 96-100). One notable contribution was J.D. can provide the advantages associated in the literature with Brown’s (1914) Subject Classification, reviewed in Beghtol both pre- and post-coordination. And a user can employ (2004). The idea of classifying by phenomena has been ap- common grammar in a query and be guided to works that plied in detail in recent years not only in the BCC but also make an argument of similar construction. in the Integrative Levels Classification (2004). Arguments that are more specific were made in other pa- Phenomena within BCC are classified within seventeen pers. Szostak (2008) argued that domain analysis—which broad classes. These reflect the idea of “integrative levels” 236 Knowl. Org. 47(2020)No.3 R. Szostak. Basic Concepts Classification (BCC)

(Kleineberg 2017); phenomena of similar levels of complex- classificationist need not and should not take sides on theo- ity should be grouped together. There is thus a class of retical disputes within a field but should rather seek a struc- “waves and particles,” another of “molecules and elements,” ture that has a place for any phenomenon identified either two classes of complex natural objects, “rocks” (which in- theoretically or empirically in the field. cludes a variety of earthly substances) and “celestial objects,” Each class is indicated notationally by a capital letter. a class of “non-human environment” that encompasses Where possible, these are the first letter of the class name: things like mountains and lakes, a class of basic “biological M for molecules and elements, R for rocks, C for culture, I entities,” a class of “flora and fauna,” two classes that capture for individual differences, and so on—but Z for celestial ob- human nature, “genetic predisposition,” and “individual jects and X for mathematical concepts since C and M had differences,” and several classes that capture elements of hu- been employed elsewhere. The first level of subdivision man society: “culture,” “art,” “politics,” “economy,” “social within each class is usually indicated by a second capital let- structure,” “technology and science,” and “health and pop- ter. Thus, CV represents values within the class of culture. ulation.” There is also a class of “mathematical concepts.” Two exceptions have been made. The single letter T indi- These seventeen broad classes are broadly similar to the cates the main subclass within “technology and science.” twenty-six classes in the Integrative Levels Classification These are among the most-used concepts in the classifica- (http://www.iskoi.org/ilc/1/ilc.php), though different ter- tion, capturing elements of books, articles, and texts more minology is employed, and the BCC combines some ILC generally. Likewise, the single letter N denotes the main classes. In both cases the goal is to be exhaustive, designating class of “non-human environment,” which includes com- one class into which each thing of which humans are aware mon indicators of place and time. can fit. The next level of subdivision employs numbers as nota- These classes were developed logically. There is some re- tion. Lower-case letters indicate further levels of subdivi- semblance between classes and the subject matter of partic- sion. CV1a is thus “ambition,” within the subclass (1. ular academic disciplines only because there is some logic to “goals”) of the subclass cultural values. It is thus easy to tell the structure of academic disciplines. But there is purposely notationally what level of hierarchy one is dealing with. no attempt to adhere to disciplinary boundaries; culture is Note that we employ the first letter of subclasses—”a” for addressed in one coherent class in BCC even though it is ambition—whenever possible. studied by several disciplines. A mixture of deduction and Subdivision usually occurs in terms of “types of X.” Val- induction was used to identify subclasses. It was often fairly ues are a type of cultural attitude, goals are a type of value, straightforward to identify these, in studying the economy and ambition is a type of goal. There are occasional excep- one naturally is curious about such things as wages, prices, tions where subdivision occurs in terms of “elements of X.” and unemployment rates. In some cases, classifications em- These are clearly indicated. For example, “providence” is an ployed in a particular scholarly field were adapted for the element of CR, “religions.” BCC. While there are competing classifications of person- As noted elsewhere, it was generally found that subdivi- ality dimensions it was fairly easy to produce a classification sion could follow logical principles. Literary warrant was for BCC that had a place for each of the dimensions empha- pursued to ensure that any concept encountered in the lit- sized in the major competing approaches. This represents a erature found a place in the schedules. However, reliance on general strategy: to provide a logical structure in which all a synthetic approach to constructing complex subject phenomena that are identified in the literature have a place. strings, and the existence of separate schedules of relators In a few cases, such as psychological disorders, it is more dif- and properties, meant that there was generally a clear logical ficult to identify a classification that will not become out- place for any concept found in the literature (or found in dated as psychological understanding changes. other classifications; we talk about comparisons of BCC The approach taken in identifying subclasses is similar to and UDC below, Szostak 2011 reported an extensive exer- the approach recommended for “domain analysis” within cise of translating DDC classes and ICONCLASS into knowledge organization (Hjørland 2017), which involves BCC; see https://sites.google.com/a/ualberta.ca/rick-szos- building a logical structure grounded on expert advice from tak/publications/ddc-to-bcc-translation-table-web-version- within the domain. As was argued in Szostak (2010), and in 2013). As the BCC is applied to collections of documents Szostak, Gnoli, and Lopez-Huertas (2016, chapter 6), the or objects, literary warrant can guide decisions as to whether procedures of domain analysis can be employed in pursuit further subdivision of subclasses is desirable. Note, though, of a general classification. It has generally not been difficult that the synthetic nature of BCC subject strings generally to render expert advice in terminology (i.e., basic concepts) allows a very precise classification of a document or object that will have a broadly shared understanding—good with recourse to flat hierarchies of phenomena. enough for the purposes of guiding diverse users to docu- Since BCC hierarchies are fairly flat, the notation for in- ments or objects—outside of the field of the expert. The dividual concepts is generally short; rarely are more than Knowl. Org. 47(2020)No.3 237 R. Szostak. Basic Concepts Classification (BCC) four or five notational spaces involved. This means that it is logical,” “physical,” or “intentional” notationally but in possible to synthesize multiple concepts. each case indicating the three main subsets of these. There is also a small set of five “changes within a phenom- 2.2 The classification of relators enon.” These are sufficiently distinct from the causal relators between phenomena above that they are given distinct nota- As with phenomena, there is a tradition within knowledge tions: “↑” for “increase” or “growth” or “development” (note organization of seeking to classify relators. Key works in- that biological development is treated separately, because it include Farradane (1967) and Perreault (1994). Szostak volves qualitative changes of such a magnitude that organisms (2012) discusses this tradition. None of these previous ef- become different “things”), “↓” for “decline,” “↕” for “fluctu- forts were successfully incorporated into a broader classifi- ations” or “cycles” or “alternation,” “∆” for “change” in gen- cation. The approach of the BCC differs from preceding ef- eral (intended to be linked synthetically to specific types of forts in stressing causal relators: relators that indicate some change), and “∩” for “stability of.” sort of influence of one thing on another (or itself). One in- There are also some nineteen non-causal relators. Many of sight guiding the BCC is the recognition that most works these are familiar in other classifications. Where possible, the have some sort of “causal” argument wherein one or more BCC maintains notation already common within knowledge phenomena exert some influence on one or more other phe- organization: “-” means “of” and “/” means “by” or “from.” nomena: (dogs)(bite)(mail carriers). A couple dozen non- Round brackets signal “type of:” Y(X) means Y of type X. causal relators are also treated and each given a notation that There are several more novel relators: “>“ is “in,” “!” is is neither a number nor letter: “\” represents “for.” “about,” “:” is “from the perspective of,” “~” is “compared The classification of causal relators relies on synthesis to to,” and “^” is “associated with” or “connected to.” an even greater degree than other elements of the BCC. A Note that all relators of all types are signaled by symbols few dozen relators are identified in a small set of very flat that are neither numbers or letters. It is thus clear in any sub- hierarchies. These are then combined with each other, or ject string which terms are relators. These symbols are never with phenomena or properties, in order to generate thou- used to denote phenomena or properties. sands of more complex relators. As noted above, the few dozen causal relators can be All causal relators are signaled notationally by an arrow: combined with other relators, phenomena, and things to “→” (sometimes pointing in the opposite or both direc- generate literally thousands of other relators. The relators tions to indicate the direction of influence). The arrow on generated through synthesis are usually more precise. Two its own can serve to indicate influence in general for cases frequent forms of synthesis should be mentioned first. We where it is specified that one phenomenon influences an- can indicate the opposite of any verb by underlining the no- other but not how. A set of more specific causal relators is tation; we thus need not specify “destroy” but merely un- then identified; each of these is given a notation of two derline the notation for “create.” Similarly, we can combine lower-case letters: “→ma” is “assembling.” There are three most relators with the notation for “again” (N2w) from the sets of “physical influences”: mechanical (eight subclasses main class N in the schedules of phenomena to capture the denoted by a first letter “m”), non-mechanical (eight sub- whole set of relators that start with the letters “re-.” classes denoted by a first letter “n”), and the four forces The most obvious form of synthesis involves combining identified in nuclear physics as acting among sub-atomic two or more causal relators. We can generate “persuade” by particles (denoted by first letter “p”). There are three sets of combining “control” and “talk.” Likewise, “force” com- “biological influences”: evolutionary (four, denoted “e”), bines “control” and “move.” “Declare” is “decide” plus developmental (four processes critical to the development “talk.” Dozens of similar combinations are listed on the of organisms, denoted “d”), and a set focused on processes BCC website but classifiers could generate others. In all unique to at least some animals (eight, denoted “a”). There cases, the notation simply combines the letters in the nota- are three sets of “intentional influences” that can only be set tions of the combined relators: “persuade” is “→rsrt” be- in motion by a sentient being (even if they only influence cause “→rs” is “control” and “→rt” is “talk.” themselves): influences that operate at the level of individu- Causal relators can also be combined with non-causal re- als (nineteen, such as believing “[→ib],” that are denoted by lators. “attach” is (cause)(to be connected) or “→^” (note “i”), influences that operate between individuals (ten, each that we generally can ignore the phrase “to be” in synthesiz- starting with “r” for relationship), and three of a spiritual ing). “Assemble” is (create)(plus)(connect) or “→gc+^.” nature that are not easily captured in the other two sets (and Causal relators can also be combined with changes within a are denoted by “s”). There is also a set of general relators that phenomenon. “Maintain” is (cause)(Stability) or “→∩.” transcend the three types of influence just mentioned (de- “Enhance” is (cause)(growth) or “→↑.” noted by “g” and including “→ga” for “allowing”). Note There are even greater possibilities for creating causal re- that we have saved notational space by not indicating “bio- lators by combining relators and phenomena. “Injure” is 238 Knowl. Org. 47(2020)No.3 R. Szostak. Basic Concepts Classification (BCC)

(cause)(injury). “Learn” is (achieve)(learning). “Classify” is two that describe the essence of the document, object, or (achieve)(classification)—which is quite different from idea. If not, the classifier will need to construct such a sen- (create)(classification). “Offend” is “cause resentment.” tence(s) from longer textual materials. It is worth emphasiz- The BCC website again lists dozens of possibilities but ing that the BCC can, therefore, be applied very broadly to many more could be created. any object or idea or document that can be described in a Relators can also be combined with properties to gener- sentence. ate further relators. “Suffer” is (experience)(bad). “Dam- The classifier can then translate the key elements of the age” is (cause)(damaged). “Compliment” is (talk)(compli- sentence into the terminology of the BCC choosing terms mentary). Dozens of such combinations can be found on from the schedules of phenomena, relators, and properties. the BCC website. More are possible. Finally, yet im- Such subject strings will almost always involve multiple phe- portantly, some relators can only be represented within nomena, usually at least one causal relator, and often one or causal strings. For example, “free” is (cause)(X)(not con- more adjective/adverbs. The classifier can ignore the ele- trol)(Y). “Evict” is (move)(from)(home or office). ments of a descriptive sentence (determiners and most pro- nouns) such as “the,” which are neither phenomena, rela- 2.3 The classification of properties tors, nor properties. If the sentence the classifier is drawing upon follows a typical grammatical construction, the classi- The schedules of properties are similar in format and nota- fier enters the BCC terms in the order they (or synonyms) tion to the schedules of phenomena. They are distinguished appear in the sentence. entirely by the use of the capital letter Q (for “qualities”) as The resulting subject strings resemble the sentences from the first letter in the notation for all properties. As with phe- which they emerge far more than the typical subject head- nomena, the next level of subdivision is represented by an- ings associated with existing library classification systems other capital letter and the following level by numbers. (and thus BCC strings express facets in a different order There are, at present, seventeen types of property identified than facet analysis tends to suggest). Knowledge organiza- in the schedules, and these in turn are divided into between tion professionals may thus find the subject strings unusual. four and ten types. The types of property are quite diverse: Yet there are huge advantages to the grammatical approach. aesthetic properties (QA) such as beautiful (QA1); behav- First of all, humans spend most of their lives speaking, talk- ioral properties (QB) such as anonymous (QB6); compara- ing, and reading in sentences. As a result, humans can more tive properties such as more (QC5); evaluative properties readily comprehend subject strings that read like sentences. such as popular (QE9); physical properties such as hard In an age when some public libraries have abandoned library (QP1); or properties associated with values such as wasteful classifications because users found them difficult to navi- (QV7). The schedules were developed inductively as con- gate, the ease of use of the BCC is noteworthy. A user can cepts were encountered in document descriptions, but then enter a search query in a sentence and be guided to the most properties were grouped together, and similar types of prop- similar subject string—and that subject string will make erty were added. sense to the user, because it is also in a grammatical format. Hierarchies of properties are very flat. Notational length Moreover, the subject string captures the essence of the doc- thus never extends beyond two upper-case letters and a ument, object, or idea in question. It deserves to be stressed number. In some cases, important opposites receive sepa- again here that the key element of a document is the ideas rate designations in the schedules (both “more” and “less”) (usually in the form of causal relationships) that it contains. but in general, opposites are captured by underlining. This BCC subject strings capture key arguments. As for objects, serves to keep the schedules of manageable size. Note that a BCC subject string can capture the composition of an ob- these properties can be employed synthetically as both ad- ject, its mode of manufacture (and perhaps place and time), jectives and adverbs. its use, and its cultural meaning: The classifier can be guided by the object description in determining which ele- 2.4 Synthesizing subject strings ments to stress in a subject string. A (golden)(ceremo- nial)(axe) can be distinguished from (steel)(axe)(for)(war). Documents are described in abstracts or book descriptions. At present, it is very difficult to ascertain which museums These often include a sentence or two that describe the might possess particular artifacts; widespread use of a classi- main argument(s) made in the document. Objects in muse- fication such as BCC could facilitate such searches immeas- ums or galleries usually also have object descriptions; again, urably. Moreover, a user could move easily between search- one or two sentences often describe the key characteristics ing for documents about certain types of artifacts and the of the objects. Ideas are commonly expressed in sentences. artifacts themselves. Whether dealing with a document, object, or idea, the clas- The notation for (steel)(axe)(for)(war) is sifier will generally then have recourse to a key sentence or (TIt\→mbNMw)(MEFe)\ PI1 ↔rxgm PI1. TIt is tool. “\” Knowl. Org. 47(2020)No.3 239 R. Szostak. Basic Concepts Classification (BCC) is “for.” “→mb” is “break” or “cut.” NMw is wood. MEFe is viduals from different groups and backgrounds to use a iron, and the parentheses around MEFe sugnal “of type.” PI1 shared classification to explore the entire universe of docu- is state. “↔rxgm” is to engage in conflict in a manner that in- ments, objects, and ideas. volves physical movement. As the BCC is applied it may The BCC has been designed very much with a digital en- prove advantageous to develop simpler notation for both vironment in mind. Nevertheless, BCC subject strings can “axe” and “war” than (tool)(for)(cutting)(wood) and (state) be used if desired for shelf marks in libraries (or indeed mu- (physical conflict with)(state). BCC subject strings are usually seums). The classifier can bold the term in a subject string less complicated than this. A classifier or user will ideally be that should be prioritized for shelving purposes. Most of the guided by the search interface to know how to treat axe or war time, this will likely be the first term anyway, for a docu- or any other term; in the absence of this interface, the BCC ment about “X influences Y” will be best grouped with schedules are easy to search in order to identify appropriate other documents about X—but the classifier in a particular terminology. case may decide that it belongs with other works about Y. If There are of course multiple ways that a particular idea we treat the notations associated with relators as if they were might be expressed in words. The ability to translate a user letters or numbers then we would collocate documents that query into a relevant subject string can be hampered by dif- describe how phenomenon X influences other phenomena, ferences in both terminology and word order. Differences how it is compared to other phenomena, how it grows or in terminology are best handled by a thesaural interface that declines, and so on. can quickly suggest controlled vocabulary to a user. In the In comparisons with both DDC (Szostak 2011) and absence of such a thesaurus, the flat and logical nature of UDC (Smiraglia and Szostak 2018), BCC strings were often BCC hierarchies should facilitate the identification of con- found to provide greater precision. BCC strings tended to trolled vocabulary. contain more distinct terms than UDC counterparts did. Differences in term order can be handled by encouraging They were nevertheless similar on average in notational a standard grammatical format (recall that we want a search length, because the flat hierarchies of BCC generally yield interface that prioritizes the word order of a search query; short notations for individual terms. BCC terminology is note though that such an interface would still identify often easy to understand. A notation such as works with a different word order if an unusual word order “CV1e→PI2b” indicates to someone with a little familiar- appeared in the query.) Szostak (2017) explored the litera- ity with the system that the document in question discusses ture on grammar and suggested that classifiers could follow some sort of influence of a particular cultural value on a par- about a dozen simple rules in translating a sentence from a ticular political institution. document or object description into a BCC string that fol- Classifiers—and the institutions that employ them— lowed a recommended word order. These rules are simple can make decisions about how much detail to include in enough—place adjectives before nouns, translate interroga- subject strings. From the perspective of users, more is gen- tive and exclamatory sentences into declarative format, use erally better for they will still find a document if the subject an extra pair of parentheses in the rare cases where it is not string contains more adjectives than the search query but clear whether a term is adjective or adverb, and so on—that can increase precision with a very precise query. But longer computers could potentially follow these rules, translating subject strings will be somewhat more costly to produce— both user queries and document or object descriptions into though this cost may be small if there is an appropriate the- a standard grammatical format. saural and grammatical interface. In museums, especially, We discussed above how BCC relies on “basic concepts” some institutions may decide to have shorter subject strings for which there are broadly shared understandings across in- than others. Classifiers would then be more selective in dividuals and groups. One key insight of the literature in the translating sentences from an object description into a subfield of semantics is that sentences that follow a common ject string. If that museum has artifacts that are unique in grammatical format serve to reduce the ambiguity associ- particular ways, short subject strings may fail to communi- ated with individual terms in that sentence. We have seen cate that uniqueness to users, including users with very pre- examples of that above: the word “axe” on its own describes cise queries. a set of artifacts that may differ in important ways; the place- It is worth noting in closing that sentences are also syn- ment of “axe” within a sentence or sentence fragment clari- thetic constructs. We do not dictate the entire set of sen- fies considerably the meaning attached to “axe” in a partic- tences that humans can utter but rather allow humans to ular context. The BCC thus tackles the challenge of termi- combine words to create any ideas they desire. We are able nological ambiguity in two complementary ways; it relies to achieve innovations in many fields of human endeavor wherever possible on basic concepts and then places these in while only rarely adding words to any human language. It is subject strings that further clarify their meaning. Such an expected that the synthetic approach taken with BCC will approach may be the best possible means of allowing indi- likewise allow the expression of diverse ideas while only 240 Knowl. Org. 47(2020)No.3 R. Szostak. Basic Concepts Classification (BCC) rarely requiring additions to its schedules (some adjust- theories and methods employed. If this were commonly ments may nevertheless be necessary if common under- done, users could then search for applications of particular standings of some terms used in BCC change over time). theories or methods. The semantic web relies on RDF triples of the form (sub- 3.0 Advantages of the BCC ject)(predicate or property)(object). That is, the semantic web relies on resources being coded in terms of synthetic In addition to several advantages noted in passing above, the combinations of phenomena, verbs, and properties. As a re- following advantages of the BCC could be stressed. It has sult, the BCC is potentially well-suited to use on the seman- been found that the BCC provides very precise descriptors tic web. Szostak et al. (2018) are exploring this connection of documents. As noted above, classifiers can capture the empirically. key elements of a document description by combining phe- Interdisciplinarity was the original motive for the BCC. nomena, relators, and properties synthetically. Smiraglia Interdisciplinary scholars most often want to investigate the and Szostak (2018) compared dozens of subject strings be- effects of phenomena studied in one discipline on phenom- tween the BCC and the Universal Decimal Classification, ena studied in another. The BCC facilitates both the search finding the former to be more precise but of similar nota- for relevant documents and the communication of research tional length. results to diverse scholars who might be interested in the As noted above, the terminology employed in the BCC same relationship. Since the BCC provides detailed classifi- reduces terminological ambiguity. Scholars of semantics cations of scholarly theories and methods, it can also poten- note that sentences serve to clarify further the meaning of tially aid scholars interested in borrowing theories and terms employed in the sentence; sentence-like subject methods from other disciplines. Note that general users are strings thus further clarify meaning. The clarity of both often also interdisciplinary in orientation and wish to dis- basic concepts and subject strings should facilitate the trans- cover relationships between phenomena without regard to lation of the BCC into languages other than English. the disciplines that may study these. As noted above, a user query in the form of a sentence can potentially be translated directly into a relevant subject 4.0 Further developments string. A user performing an exploratory search might wish to The Basic Concepts Classification is still being developed. make subtle changes to a search query. A search interface In particular, the schedules of natural science phenomena employing BCC could alert users to possibilities. For exam- are slowly being expanded. In recent research, Smiraglia and ple, a user searching for (dogs)(biting)(mail carriers) could Szostak (2018) have compared subject headings in the Uni- be alerted to documents addressing (cats)(biting)(mail car- versal Decimal Classification with those in BCC. The vast riers) or (dogs)(licking)(mail carriers) or (dogs)(bit- majority of the time the BCC schedules have already con- ing)(neighbors). The user can thus easily follow their curi- tained the concepts needed to capture the meaning of a osity to a host of related subjects. This is much harder to do UDC subject heading (often with greater precision). In a within precoordinated classifications. Note that the user minority of cases, the BCC schedules have been expanded can choose to alter nouns or verbs or adjectives/adverbs in to facilitate comparison. The BCC has thus been developed their searches. The BCC thus instantiates a “web-of-rela- to a point where it can be applied to collections of docu- tions” approach that allows users to find information re- ments and/or objects. Further attempts to apply or compare lated in a host of ways to the initial query. the BCC will undoubtedly result in further smallish addi- The BCC is thus compatible with innovative visualiza- tions to the schedules. Suggestions regarding additions are tion techniques. An interface could allow users to experi- most welcome. ment with various sorts of changes to an original search The developer of the BCC believes that there is enough query—perhaps by simply sliding a mouse over the differ- consensus—for the purposes of BCC—in most scholarly ent elements of the query. It could also guide users to related fields on how best to classify the things they study. The ease material; for example, from searching (dogs)(bite)(mail car- with which the BCC can be applied across diverse fields riers) to (mail carriers)(go to)(hospitals). provides empirical support for this hypothesis. In some Users and search interfaces could likewise easily move be- fields, though, such as the classification of psychological tween broader and narrower terms in the hierarchies of phe- disorders, it is far less clear what the best approach is. In such nomena (or properties). They could similarly move between cases the BCC website indicates that there is a challenge, simple relators and more complex relators formed via syn- and users are invited to suggest a path forward. The devel- thesis. oper of BCC will also interact with field experts. In partic- The BCC classifies scholarly theories and methods in ular, Smiraglia and Szostak are planning to gather experts in some detail. It is thus possible to classify works in terms of music classification to improve the BCC treatment of mu- Knowl. Org. 47(2020)No.3 241 R. Szostak. Basic Concepts Classification (BCC)

‘sical genres in particular and music more generally. One 5.0 Limitations and criticisms goal of this encyclopedia is to discuss best practices for classification in diverse fields, and the insights of other articles The BCC needs to be fleshed out in certain respects. Yet it in this encyclopedia will naturally be incorporated into the also needs to be applied; only in application can the feasibil- BCC. Most of the elements necessary for a classification of ity and desirability of the classification be adequately as- authorial perspective (such as ethical and ideological atti- sessed. It has been conjectured above that the BCC has tudes) already exist within BCC. Others identified in Szos- many advantages, and these have been borne out in small- tak (2015b) will be added. scale applications to dozens of objects or documents, but The BCC is being translated into linked open data. As the full advantages and limitations of the BCC can only be part of an international research grant, it and the UDC will assessed in a larger-scale application. This is not an uncom- be compared to the terminology employed at present in the mon situation; classification systems tend naturally to be re- LOD cloud (Szostak et al. 2018). The promise of the seman- vised as they are applied. Nevertheless, there are challenges tic web can only be realized if there is at least interoperability in simultaneously developing and applying a classification. across the terminology employed in different online re- The BCC has been developed by one person, albeit with sources. It is hoped that the BCC and/or UDC can be use- a great deal of advice from others. It lacks the bureaucratic ful in encouraging interoperability or use of a common vo- support that some other classifications possess. These chal- cabulary. The comparison of BCC terminology with the lenges may be best addressed as the system is applied to large LOD cloud will also further identify areas in which the collections of objects or documents. BCC schedules should be extended. We have touched above on various possible critiques of the A better interface will also be developed to allow the BCC BCC. Most obviously, is it true that the concepts employed to be more readily applied to collections of ideas, documents, in the BCC are really basic concepts? The BCC has now been or objects. User testing can then be pursued to empirically as- developed to a degree that this hypothesis can be tested em- sess the usefulness of BCC in providing access to collections. pirically. Does the structure of the BCC somehow penalize A prototype interface that prioritizes the order in which documents that pursue theories that might organize the search terms are entered has already been developed, and world differently? Though the BCC strives to find a logical work is proceeding on a thesaural interface. Though there are place for all concepts, this question can really only be evalu- programming challenges in developing the sort of interface ated empirically as the BCC is applied to a large and diverse that has been mentioned at several points above, there does corpus of documents. Recall, though, that the standard by not appear to be any insurmountable barrier. which the BCC should be evaluated is not perfection but The fields of information retrieval and knowledge organi- whether it performs better than other classifications. zation have developed separately in recent decades. Glushko The BCC unfortunately follows the practice of most of (2013) is one of many authors who urges a reconciliation. He, the world’s general classifications in not providing detailed like others, appreciates two huge challenges in standard ap- notes on which sources were used in developing the classifi- proaches to information retrieval: 1) different texts employ cation. For example, which psychologists were consulted different terminology for the same idea; and, 2) complex ideas (and why) in developing the classification of personality di- are not well captured by searches for independent “bags of mensions? Such notes would allow others to better evaluate words.” The interface envisioned here addresses both of these the system and in particular to appreciate whether it reflects concerns; it will ideally translate a search query not just into the latest thinking among scholars. Such notes might also controlled vocabulary but controlled vocabulary structured signal possible gaps or biases in the classification. The au- grammatically to match subject strings developed using thor appreciates that it would be useful to add such notes in BCC. It then will guide users to make small adjustments to the future. In the meantime, the application of the BCC to their query if they wish. It thus promises to achieve far greater a wide range of (especially recent) documents and objects precision in search than existing information retrieval tech- can serve to identify any necessary additions to the sched- niques are capable of (see Hjørland, 2012, for the general ar- ules; only very rarely in the developer’s experience has there gument that subject classification can achieve greater preci- not been an obvious place to put such additions. sion than information retrieval techniques alone). Ideally, the BCC would be accompanied by a thesaurus Notes that would provide classifiers with clear advice on synonyms for BCC terminology. The reliance of the BCC on 1. Broughton (2015, 383) “facet indicator a notational basic concepts should facilitate the construction of such a symbol used to indicate a particular element of a com- thesaurus. It might be structured in a similar manner to pound subject: e.g. in DDC, 09 is used to introduce WordNet. place, and in UDC = indicates the language of the document.” See further in Broughton (2015, 323-6). 242 Knowl. Org. 47(2020)No.3 R. Szostak. Basic Concepts Classification (BCC)

2. https://sites.google.com/a/ualberta.ca/rick-szostak/re- cessing and Management 49: 545-57. doi:10.1016/j.ipm. search/basic-concepts-classification-web-version- 2012.10.001 2013/guiding-principles Hjørland, Birger. 2017. “Domain Analysis.” Knowledge Or- 3. Broughton (2015, 376): “Bliss’s Bibliographic Classifica- ganization 44: 436-64. tion is a scheme often regarded as the most scholarly of “Integrative Levels Classification.” 2004-. www.iskoi.org/ the general schemes. It was never used in Bliss’s native ilc/ America, but was favoured by a number of UK and Julien, Charles-Antoine, Pierre Tirilly, Jesse Dinneen and Commonwealth academic and special libraries. Because Catherine Guastavino. 2013. “Reducing Subject-Tree of its unique main class for social welfare, the first edi- Browsing Complexity.” Journal of the American Society tion, BC1, was widely adopted by many charity and so- for Information Science & Technology 64: 2201-23. cial welfare libraries in the UK. The second edition, BC2, Kleineberg, Michael. 2017. “Integrative Levels.” Knowledge is the only general scheme built on faceted principles Organization 44: 349-79. published in the Western world.” Mazzocchi, Fulvio, Melissa Tiberi, Barbara De Santis and Paolo Plini. 2007. “Relational Semantics in Thesauri: References Some Remarks at Theoretical and Practical Levels.” Knowledge Organization 34: 197-214. Bliss Classification Association. n.d. “The Classification Perreault, Jean M. 1994. “Categories and Relators: A New BC2.” https://www.blissclassification.org.uk/bcclass. Schema.” Knowledge Organization 21: 189-98. shtml Sauperl, Alenka. 2009. “Precoordination or Not? A New Beghtol, Clare. 2004. “Exploring New Approaches to the View of the Old Question.” Journal of Documentation Organization of Knowledge: The Subject Classification 65: 817- 33. of James Duff Brown.” Library Trends 52, no. 4: 702-18. Smiraglia, Richard. 2001 The Nature of “a Work;” Implica- Broughton, Vanda. 2015. Essential Classification. 2nd ed. tions for the Organization of Knowledge. Lanham, MD: London: Facet Publishing Scarecrow Press. Brown, James Duff. 1914. Subject Classification, with Ta- Smiraglia, Richard and Rick Szostak. 2018. “Converting bles, Indexes, etc., for the Sub-division of Subjects. 2nd rev. UDC to BCC: Comparative Approaches to Interdiscipli- ed. London: Grafton. narity” In Challenges and Opportunities for Knowledge Classification Research Group. 1969. Classification and In- Organization in the Digital Age: Proceedings of the Fif- formation Control. London: Library Association. teenth International ISKO Conference, 9-11 July 2018 Farradane, Jason. 1967. “Concept Organization for Infor- Porto, Portugal, ed. Fernanda Ribeiro and Maria Elisa Cer- mation Retrieval.” Information Storage and Retrieval 3: veira. Advances in Knowledge Organization 16. Baden- 297-314. Baden: Ergon, 530-8. Fox, Melodie J. DATE “Review of Szostak 2003 and 2004.” Szostak, Rick. 2003. A Schema for Unifying Human Science: Knowledge Organization 39: 300-3. Interdisciplinary Perspectives on Culture. Selinsgrove, Glushko, Robert J., ed. 2013. The Discipline of Organizing. PA: Susquehanna University Press. Cambridge: MIT Press. Szostak, Rick. 2004. Classifying Science: Phenomena, Data, Gnoli, Claudio. 2016. “Classifying Phenomena. Part 1: Di- Theory, Method, Practice. Dordrecht: Springer. mensions.” Knowledge Organization 43: 403-15 Szostak, Rick. 2008. “Classification, Interdisciplinarity and Gnoli, Claudio. 2017a. “Classifying Phenomena. Part 2: the Study of Science.” Journal of Documentation 64: 319- Types and Levels.” Knowledge Organization 44: 37-54. 32. Gnoli, Claudio. 2017b. “Classifying Phenomena. Part 3: Fac- Szostak, Rick. 2010. “Universal and Domain-specific Clas- ets.” In Dimensions of Knowledge: Facets for Knowledge sifications from an Interdisciplinary Perspective.” In Organization, ed. Richard Smiraglia and Hur-Li Lee. Paradigms and Conceptual Systems in Knowledge Organ- Würzburg: Ergon, 55-67. ization: Proceedings of the 2010 conference of the Interna- Gnoli, Claudio. 2018. “Classifying Phenomena. Part 4: tional Society for Knowledge Organization Rome, Febru- Themes and Rhemes.” Knowledge Organization 45: 43- ary, 2010, ed. Claudio Gnoli and Fulvio Mazzochi. Ad- 53. vances in Knowledge Organization 12. Wurzburg: Er- Hjørland, Birger. 2012. “Is Classification Necessary After gon, 71-7. Google?” Journal of Documentation 68: 299-317. doi: Szostak, Rick. 2011. “Complex Concepts into Basic Con- 10.1108/00220411211225557 cepts” Journal of the American Society for Information Hjørland, Birger. 2013. “Facet Analysis: The Logical Ap- Science & Technology 62: 2247-65. proach to Knowledge Organization.” Information Pro- Szostak, Rick. 2012. “Classifying Relationships” Knowledge Organization 39: 165-78. Knowl. Org. 47(2020)No.3 243 R. Szostak. Basic Concepts Classification (BCC)

Szostak, Rick. 2014a. “Classifying for Social Diversity” Conference Rio de Janeiro, Brazil, September 27-29, 2016, Knowledge Organization 41: 160-70. eds. J. A. Chaves Guimarães, S. Oliveira Milani and V. Szostak, Rick. 2014b. “Skepticism and Knowledge Organi- Dodebei. Advances in Knowledge Organization 15. zation” In Knowledge Organization in the 21st Century: Würzburg: Ergon, 359-67. Between Historical Patterns and Future Prospects; Pro- Szostak, Rick. 2017. “A Grammatical Approach to Subject ceedings of the 2014 Conference of the International Soci- Classification in Museums.” Knowledge Organization ety for Knowledge Organization Krakow, May, 2014, ed. 44: 494-505. Weislaw Babik. Advances in Knowledge Organization Szostak, Rick, Claudio Gnoli and Maria López-Huertas. 14. Wurzburg: Ergon, 57-64. 2016. Interdisciplinary Knowledge Organization. Berlin: Szostak, Rick. 2015a. “A Pluralistic Approach to the Philos- Springer. ophy of Classification” Library Trends 63: 591-614. Szostak, Rick, Andrea Scharnhorst, Wouter Beek and Rich- Szostak, Rick. 2015b. “Classifying Authorial Perspective.” ard Smiraglia. 2018. “Connecting KOSs and the LOD Knowledge Organization 42: 499-507. Cloud.” In Challenges and Opportunities for Knowledge Szostak, Rick. 2016. “Employing a Synthetic Approach to Organization in the Digital Age: Proceedings of the Fif- Subject Classification across Galleries, Libraries, Ar- teenth International ISKO Conference, 9-11 July 2018 chives and Museums.” In Knowledge Organization for a Porto, Portugal, ed. Fernanda Ribeiro and Maria Elisa Sustainable World: Challenges and Perspectives for Cul- Cerveira. Advances in Knowledge Organization 16. Ba- tural, Scientific and Technological Sharing in a Connected den-Baden: Ergon, 521-9. Society. Proceedings of the Fourteenth International ISKO

244 Knowl. Org. 47(2020)No.3 B. Hammarfelt. Discipline

Discipline†

Björn Hammarfelt

University of Borås, Swedish School of Library and Information Science, Allégatan 1, Borås, Sweden,

Björn Hammarfelt is an associate professor at the Swedish School of Library and Information Science (SSLIS), University of Borås. His research is situated at the intersection of information studies and sociology of science, with a focus on the organisation, communication and evaluation of research. His most recent work has mainly been focused on how scholars use and respond to bibliometric measurement.

Hammarfelt, Björn. 2020. “Discipline.” Knowledge Organization 47(3): 244-256. 71 references. DOI:10.5771/ 0943-7444-2020-3-244.

Abstract: “Discipline” is commonly used to denote particular areas of knowledge, research and education. Yet, the concept is often not very well defined or even explicitly discussed when used in knowledge organisation and related fields. The aim of this article is to encourage and facilitate further reflections on academic disciplines, while at the same time offering insights on how this elusive concept might be understood. An overarching argument is that “discipline” should foremost be understood in relation to institutional and organisational features, and this is what distinguishes it from related terms such as, field, domain or topic. The etymology and history of the concept are reviewed along with a discussion of attempts to define and conceptualise disciplines. Insights are offered on how disciplines might be studied. Regardless of our views of disciplines, either as inherently out-dated constructs or as important features of a well-functioning academia, it is concluded that further precision or care in explicating the concept is needed.

Received: 9 September 2019; Revised: 30 September 2019; Accepted: 26 October 2019

Keywords: discipline, disciplines, research, knowledge organization, knowledge areas, academic fields

† Derived from the article of similar title in the ISKO Encyclopedia of Knowledge Organization Version 1.0 published 2019-09-04. Article category: Theoretical concepts. This paper partly extends and develops arguments made in a recent conference paper, “What is a Discipline? The Conceptualization of Research Areas and their Operationalization in Bibliometric Research” (Hammarfelt 2018), which in turn builds on the author’s PhD dissertation (Hammarfelt 2012). Muhebera Bizimana, Helena Francke, Frances Hultgren, Jenny Johannisson, Jan No- lin, Ola Pilerot, Veronica Johansson, Fredrik Åström and two anonymous reviewers have generously commented on earlier drafts of the manuscript. The author is grateful for encouragement and insightful suggestions from the editor of the ISKO Encyclopedia of Knowledge Organization, Birger Hjørland.

1.0 Introduction Within library classification, “disciplines” have often been used as the fundamental principle of organization. “Discipline” is a key concept in knowledge organisation The Dewey Decimal Classification (DDC), for example, (KO) and related fields, yet it is often not very well under- states (Dewey 1979, xxxi; emphasis added): “a work on wa- stood or defined. It is commonly used to define and deline- ter may be classed with many disciplines, such as metaphys- ate knowledge areas, often when referring to academic fields ics, religion, economics, commerce, physics, chemistry, ge- of research and education. Clearly, it is a central concept in ology, oceanography, meteorology, and history. No other fea- the broad understanding of KO as the study of “the concep- ture of the DDC is more basic than this: that it scatters subjects tual systems, the social fields, and the activity system of by discipline.” Classifying by disciplines is also called “aspect knowledge” (Hjørland, 2016). Yet, discipline is also im- classification.”1 portant in relation to what Hjørland (2016) describes as a Similarly, library and information science scholars as well more precise definition of KO; the study of specific as bibliometricians often use the term discipline to depict knowledge organising systems in terms of the organisation and separate knowledge, institutional structures, researchers of bibliographical databases and libraries. and resources; “disciplines have been a standard framework for bibliometric analyses of bodies of literature and studies

Knowl. Org. 47(2020)No.3 245 B. Hammarfelt. Discipline of scholarly communication and user communities” (Palmer ence.com 2019). In ancient Rome, “disciplina” was a minor and Cragin 2008, 172). However, “discipline” is usually not goddess who represented martial discipline. She was primar- very well defined, if defined at all. The aim of this article is ily worshipped by soldiers and is likely to have been a creation to encourage and facilitate further reflection on how this of Hadrian (Encyclopedia Mythica 2019). concept is used and to contribute to discussions regarding A German etymological dictionary (Kluge [1891] 2019), how “discipline” can be defined, operationalised and stud- on the other hand, gives the following explanation of the ied. I build on an ongoing discussion about how academic origin of the German “Disziplin:” Disziplin comes from disciplines should be understood and more broadly defined, “disciplina,” which in turn is derived from “discipulus.” and draw on literature from a range of fields. An important Discipulus in turn relates to “discipere,” where the latter is point of departure for the paper is that discipline points to marked as a deduced form (“erschlossen”) meaning to grasp institutional and organisational characteristics such as de- (“erfassen”). A similar explanation is offered by Stichweh partments, conferences and labour markets that distinguish (1992, 4) who describes the term’s etymology in the follow- it from concepts such as field, domain or topic. ing manner: The article is structured as follows: first the etymology and definitions of “academic discipline” are given and the The term ‘discipline,’ deriving from the Latin discere, historical roots of the concept are discussed. This back- was of course already known in the early Middle Ages ground offers a vantage point from which the future of dis- (Marrou 1934; cf. Evans 1980, 96-97). Until the eight- ciplines can be discussed. Thereafter, derivatives of “disci- eenth century, the history of the term disciplina was pline,” such as cross-, inter-, trans- and post-disciplinary are closely linked to the history of the term doctrina. In considered with a particular focus on what these concepts fact, the terms are frequently indistinguishable. Char- infer in connection with the concept of “discipline.” Subse- acteristically, they are used in the context of teaching quently, insights are offered into how disciplines can be the- and instruction, and refer to a systematic entirety of orised. Related concepts, such as field, domain and topic are doctrines with which a student is presented in the con- examined in light of their relation to the concept of disci- text of instruction. If the two terms are to be differen- pline. Some of the main insights derived from the broad and tiated, then doctrina refers to the teacher’s side of in- extensive literature on disciplines are summarized by outlin- struction and disciplina to the student’s side. ing a few directions for how “discipline” might be fruitfully conceptualised and understood in information studies and By establishing the etymology of the concept, we find an knowledge organisation. emphasis either on social aspects (followers) or on the content of knowledge (doctrines). This tension between disci- 2.0 Etymology pline as a concept explaining social relations and organisation, and its role in demarcating specific areas of knowledge Discipline is derived from the Latin discipulus; however the is a recurrent theme in attempts to define the concept. etymology of this word is not entirely clear and none of the existing theories fully satisfy etymologists (see De Vaan 3.0 Understanding disciplines 2008). In old English, it is supposedly derived from discipul (fem. discipula), meaning the “one who follows another for 3.1 Definitions the purpose of learning,” especially “the personal followers of Jesus Christ during his life, the twelve Apostles chosen or According to the OED (2019), the noun “discipline” has called by him to be his immediate associates,” a Biblical loan two meanings: from Latin discipulus “pupil, student, follower” (Online Et- ymology Dictionary 2019). 1) the practice of training people to obey rules or a The term discipline has a long history in English, where code of behaviour, using punishment to correct diso- the earliest use according to The Oxford English Dictionary bedience, and (OED), is recorded in 1398. Chaucer used it in relation to 2) a branch of knowledge, typically one studied in science as early as 1405: “Assaye in myn absence This disci- higher education. plyne and this crafty science” (OED 2019). A more specific use of the term in relation to areas of teaching and scientific In addition, discipline can be used as a verb in the sense: inquiry occurs in English from mid seventeenth century “train (someone) to obey rules or a code of behaviour, using and onwards. punishment to correct disobedience.” Discipline bears similar meanings in Sanskrit: “instruc- Discipline is commonly used in contexts where a distinct tion,” “education” (Monier-Williams [1851] 2019) and in chain of authority exists, such as the church or the military. Greek: “training to obey rules,” “self-control” (Wordrefer- In this sense “academic discipline” can be seen as a form of

246 Knowl. Org. 47(2020)No.3 B. Hammarfelt. Discipline specific and rigorous training that will turn out practition- 3) “theories and concepts” that can organise knowledge ers “who have been ‘disciplined’ by their discipline’ for their effectively, own good” (Krishnan 2009, 8). To Michel Foucault, “disci- 4) specific “terminologies or technical language,” pline,” in its more general sense, is part of an often brutal 5) developed particular “research methods,” political force that police certain behaviours while excluding 6) an “institutional manifestation” in the form of a those that deviate from the norm, and eventually “disci- subject taught in universities academic departments pline” is internalised by the subject itself (Foucault 1995, and professional organisations. 223): “The disciplines characterize, classify, specialize; they distribute along a scale, around a norm, hierarchize individ- An important additional point in the establishment of dis- uals in relation to one another and, if necessary, disqualify ciplines is control over specific channels for disseminating and invalidate.” knowledge (journals, book series and conferences). The im- Here we are interested in “discipline” as it is used to de- portance of communication is, for example, emphasised by note a “branch of knowledge” or more specifically, its use in Simon (2011, 1) in his study of the formation of physics labelling fields of research and education within the aca- during the second half of the nineteenth century, as such a demic system. Academic disciplines are less associated with perspective may be “an efficient way of combining the fo- violence and obedience, although power relations and cus on knowledge and practice with that on institutional, boundary keeping are certainly important aspects when de- occupational and social roles.” fining the concept. There is, in fact, a strong connection be- International recognition and generally shared norms re- tween discipline and power, as Pierre Bourdieu [1984] garding academic credibility are other characteristics that 2003) asserts in his study of the struggle for power between are often associated with academic disciplines (Becher and faculties (and disciplines). Trowler 2001, 41). Control over how knowledge is dissemi- The concept “discipline” is less straightforward or well nated and acknowledged is emphasised by Lenoir (1997, 47) defined than one might think. Disciplines could be viewed who asserts that: “disciplines are the institutional mecha- as systems that produce statements about the world, but dis- nisms for regulating the market relations between consum- ciplines are limited and restricted in themselves while the ers and producers of knowledge.” In regulating the market “discursive formation” in which they are constituted goes of knowledge, disciplines distribute status and rewards. In far beyond disciplinary boundaries (Foucault 1971, 179). this regard, disciplines are closely related to “professions” Disciplines should not, therefore, be confused with topics, (Abbott 1988), and the two are often tightly intertwined. discourses, subjects or interests; rather they should be un- Indeed, establishing an “academic discipline” is often a key derstood as knowledge institutions or knowledge systems. strategy for strengthening the autonomy and status of a pro- A discipline is partly defined by institutional structures fession. For example, Danziger (1997) described how the within departments, but “international currency is an im- discipline “psychology” annexed many fields in which psy- portant criterion, as is a general though not sharply defined chology had not made important contributions (such as set of notions of academic credibility” (Becher and Trowler personality, motivation and areas of applied psychology). 2001, 41). As formulated by Lenoir (1997, 46), disciplines Grouping these diverse areas together as branches of one embody “the infrastructure of science” through publica- discipline undoubtedly had practical advantages. It ad- tion outlets, academic conferences and associations. A fur- vanced the cause of professionalization by implying that the ther defining trait of an academic discipline is the existence more practically oriented branches had a respectable link to of undergraduate and graduate education as well as text- basic science, and it legitimized the otherwise esoteric inter- books and a core of canonical publications. ests of the academics by implying that their work had significant practical applications. But, for the most part, such im- 3.2 Characteristics plications were nothing more than promissory notes to be cashed in at some time in the future. In short, it was much Due to the ephemeral nature of disciplines, many authors easier to annex such fields institutionally than to assimilate abandon attempts to come up with exact definitions and them intellectually, and psychology, therefore, became ex- opt instead for broader conceptualisations of what a disci- tremely fragmented and incoherent. pline is and how it can be characterised. In an attempt to Sugimoto and Weingart (2015) adopt a similar approach discern the characterises of an academic discipline, Krish- to that of Krishnan in their review of central aspects of aca- nan (2009) posits six qualities. In his view, disciplines have: demic disciplines. Depending on the chosen perspective they suggest that disciplines can be discussed in terms of 1) a particular “object of research,” their cognitive, social and communicative characteristics. 2) a body of “accumulated specialist knowledge” re- Moreover, aspects such as separateness, tradition (history) ferring to their object of research, and institutional anchorage play important roles. In partic-

Knowl. Org. 47(2020)No.3 247 B. Hammarfelt. Discipline ular, Sugimoto and Weingart highlight the importance of searcher in disciplines like sociology today to master the en- “narratives” in developing a “disciplinary” identity. The im- tire knowledge of a whole discipline. Dogan states (2001, portance of a shared and reiterated history of the field is in- 14851): “The process of specialization has tended to disjoin deed an important feature of a discipline. Such histories can activities which had previously been united, and to separate take different forms, but a key event in these stories is often scholars belonging to the same formal discipline, but who the “birth” of the discipline. Sugimoto and Weingart (2015) are interested in different fields.” distinguish a few typical narratives: that of the founding fa- thers or mothers, a key event (for example a landmark con- 3.4 Disciplines as social and epistemological entities ference), or publication (a specific journal or book). How- ever, the story can also be about a movement (women’s One of the more elaborate attempts to define discipline is rights or environmental concerns) originating outside “aca- provided by Turner (2000, 47) who writes: demia.” Yet, this shared history is constantly re-written as disciplines develop, and a change of focus may result in new Disciplines are kinds of collectivities that include a histories being written. As concluded by Hjørland (1998, large proportion of persons holding degrees with the 166) in his study of the formation and classification of psy- same differentiating specialization name, which are chology: “disciplines can continue to exist and grow even if organized in part into degree-granting units that in the criteria that played a critical role in their establishment part give degree-granting positions and powers to per- are later abandoned.” sons holding these degrees; persons holding degrees of this particular specialized kind are employed in posi- 3.3 The size of disciplines tions that give degree-granting powers to them, such that there is an actual exchange of students between Jacobs (2013, 27), builds on Turner in his definition of dis- different degree-granting institutions offering degrees ciplines when stating: “A discipline is defined as a broadly in what is understood to be the same specialization. accepted field of study that is institutionalised as a degree- granting department in a large number of colleges and uni- This rather lengthy definition focuses on disciplines as or- versities.” He continues by stating (Jacobs 2013, 28) that ganisations regulating how rewards and positions are allo- disciplines are “organised groups that certify knowledge in cated, and it explains how a discipline reproduces itself. the university context.” However, the definition used by Ja- Richard Whitley (2000) provides us with a similar defini- cob differs from that of Turner in its emphasis on the size. tion of disciplines although focusing less on the formal ap- Hence, in order to be a “proper” discipline an academic paratus of degree granting. In his words (81), academic dis- field needs to have a certain size. Accordingly, Jacobs (2013, ciplines are “units of labour market control which trained 30) suggests that physics, which is represented in a vast ma- knowledge producers in particular skills that monopolised jority of colleges and universities, might be defined as a dis- contributions to particular intellectual goals.” cipline, while it can be questioned if the smaller field of as- Thus, to be defined as a discipline requires that a partic- tronomy, which is not represented at a majority of universi- ular area of research is recognised and acknowledged, not ties, is a discipline. Clearly there is a good argument to be only by researchers themselves but by outsiders as well. Sim- found here, as a certain size is a requisite for achieving the ilarly, demarcation and control are emphasised by Kohler status of being a discipline. Yet, making discipline about size (1982 quoted in Gieryn 1999, 34): “Disciplines are political bears problems. First, where do we draw the line: how large institutions that demarcate areas of academic territory, allo- must a field be to count? And second, what happens to dis- cate the privileges and responsibilities of expertise, and ciplines that are on a downward trajectory in terms of size? structure claims on resources.” If, for example, fewer institutions offer education and re- The definitions above highlight different aspect of de- search in “classical languages,” do they then risk becoming fining the concept of discipline, yet they all confer a view of “de-disciplined?” Still, size does matter when discussing dis- disciplines as social, and foremost organisational units, ra- ciplines, and this issue becomes especially noteworthy in re- ther than intellectual or epistemological ones, with a main lation to neighbouring concepts such as research field, focus on training and teaching as well as on upholding and knowledge domain and research specialty. protecting boundaries. Indeed, proponents of interdiscipli- The size and growth of disciplines have consequences for narity would argue that new knowledge could be seen as their ability to communicate. In large heterogeneous disci- threatening “disciplinarity” rather than strengthening it. plines a given member may only grasp certain aspects of the Importantly then, disciplines are not to be confused with knowledge produced. Indeed, Dogan (2001) argues that dis- topics, areas, subjects or interests; rather they should be un- ciplines are no longer the most important units in scientific derstood as social institutions or social systems, with the communication. In his view, it is impossible for any remain focus of reproducing themselves through the training

248 Knowl. Org. 47(2020)No.3 B. Hammarfelt. Discipline of new “disciples.” The heterogeneous nature of disciplines, 4.0 The history and future of disciplines due both to differences in national contexts and between branches of knowledge (social sciences, natural and medical The need to define and separate different areas of learning sciences and the humanities), makes it hard to arrive at a def- and knowledge has a long history. In western thinking, this inite definition. While recognising these challenges, it is ev- tradition can be traced back at least to Plato and Aristotle ident that the most ambitious attempts to formulate a defi- where the latter in his Lyceum divided areas of learning into nition all point to the importance of organisational features, three kinds: productive (for example engineering, strategy where control over educational degrees, communication and rhetoric), practical (politics and ethics) and theoretical channels and labour markets are key features. (physics, mathematics and theology) (Kenny and Anselm On a more epistemological level, the formation of disci- 2019). Later, the “seven liberal arts,” which were established plines can be related to how “paradigms” are formed in in ancient Rome, formed the basis for separating teaching Kuhn’s theory on the structure of science (2012). “Disci- areas in the medieval universities and onwards. The liberal pline” is not explicitly used by Kuhn, rather he used the arts grew out of several traditions where in particular the in- term “scientific community.” Yet, in his 1969 postscript, fluence of Islamic learning has been emphasised in more re- Kuhn introduces the concept of “disciplinary” matrix in or- cent scholarship (Kimball 2010: 1-12) The seven liberal arts, der to explain and clarify his use of the term paradigm. A “trivium”(grammar, rhetoric and logic) and “quadrivium” key element of the disciplinary matrix is, according to Kuhn (geometry, arithmetic, music and astronomy) shaped educa- (186), the “concrete problem-solutions” or “exemplars” that tional institutions for a considerable time, and the tradition students encounter in their education. Kuhn emphasises of “liberal arts” is still strong today, especially in an Ameri- the learning of conventions and practices as an important can tradition. Similar to Aristotle’s theoretical sciences, the feature in the formation of a disciplinary matrix, yet he also liberal arts are foremost viewed as part of an effort to be- emphasises that this training becomes less relevant when re- come educated in a general sense (“bildung”), rather than to search becomes more specialised: “As their training devel- acquire specialist knowledge. The establishment of disci- ops, however, the symbolic generalisations they share are in- plines can be seen as continuing the tradition of “artes” and creasingly illustrated by different exemplars.” In conclusion, several fields of learning, such as literature and physics, are what Kuhn calls “paradigm” or in later writings “discipli- also considered to be contemporary disciplines. However, nary matrix” is important to consider when discussing the the emergence of disciplines can also be seen as a distinct nature of disciplines, not least as many subsequent theories shift from the generic to the specific, and from the educa- on “disciplines” and “fields” are directly inspired by the tion of “full citizens” to the training of specialists. Moreo- Kuhnian view (for example, Whitley 2000; Becher and ver, disciplines are tightly connected to the emergence of the Trowler 2001). However, it is important to emphasise that modern research university, and the idea of teaching and re- “academic disciplines” often contain several “paradigms” search as the two main missions of scholars. In contrast, the (multi-paradigmatic fields) or they may be viewed, as is the liberal arts have mainly been associated with the teaching of case with many disciplines in the humanities and social sci- a tradition rather than the production of new knowledge. ences, as “pre-paradigmatic.” This means that several competing “ways of doing” research exists (multi-paradigmatic), 4.1 The emergence of disciplines or it might be that few agreed-upon conventions and practices exist thus making the field pre- or non-paradigmatic. Academic disciplines, in their more modern form, were first Importantly, disciplines change and develop over time. developed in Germany during the eighteenth and early nine- Toulmin (1977) suggested that a science [discipline] devel- teenth century. Notably, the term “discipline” was used long ops continually as either the content or the institution will before the eighteenth century as Stichweh (2001, 13727) remain stable as the other changes; if a discipline encounters notes: “There exists a long semantic prehistory of disciplina a theoretical crisis, its institution will insure its survival for a as a term for the ordering of knowledge for the purposes of time until a new theoretical foundation has been estab- instruction in schools and universities. But only the nine- lished. If its institutional existence is threatened, its theoret- teenth century established real disciplinary communication ical contributions may survive in a new organisational set- systems.” Hence, the term “discipline” has a much longer ting. Moreover, loosely defined fields may gradually de- history than the modern concept of academic disciplines as velop into more stable disciplines. Library and information relatively well-defined fields of knowledge production. science, for example, started as a multidisciplinary field In explaining the emergence of disciplines, Stichweh based on literary studies, children’s culture studies, psychol- (1992) points to three related developments: ogy, sociology, management, computer science, etc., and then developed towards a monodiscipline in its own right (Tengström 1993, 12).

Knowl. Org. 47(2020)No.3 249 B. Hammarfelt. Discipline

1) the stabilisation of scientific communities and the more than hundred years old (Klein 1996, 6). Conse- establishment of formal organisational structures in quently, it should be remembered that disciplines refer to a universities, specific organisation of knowledge production that can be 2) the separation of disciplines from recognised pro- situated historically, and while this organisation principle fessions like law, theology and medicine, and has dominated knowledge production over the last hundred 3) the development of a formal and well functioning years, it is by no means given that they will play the same role scholarly communication system. in the future.

The importance of communication structures is empha- 4.2 Challenges to academic disciplines sised also by Bawden (2017) in his description of the development of chemistry: “Chemistry was in the lead in the de- The organisation of academic research has gone through velopment of disciplinary speciality generally, with its asso- profound changes in the late twentieth century, and the im- ciated science communication system, in the latter part of portance and relevance of “disciplines” has been challenged the nineteenth century, with initiatives including learned with developments described under labels such as, “mode 2 societies, specialist libraries, conferences and their proceed- knowledge production” (Gibbons et al. 2001) and “post ings, journals, monographs, abstracts, reviews, and guides normal science” (Ravetz 1999). According to these authors, to the literature.” science is now entering a phase where knowledge produc- Similar explanations are provided by Whitley (2000, 57): tion is thoroughly integrated in society, and in which the “By systematically connecting organisational status and au- control of science is no longer only in the hands of a disci- thority to extra-local reputations for contributions to collec- plinary elite, but involves also an “extended peer commu- tive intellectual goals, the nineteenth- century university sys- nity.” The terms “post-normal” signals that this is not what tem bureaucratized intellectual production and organized Kuhn defines as “normal science” but rather a type of into distinct, specialized disciplines.” Similarly, Clark (2006, knowledge production in which “facts are uncertain, values 55), in his study of the establishment of the modern univer- in dispute, stakes high, and decisions urgent” (Ravetz 1999, sity, exemplifies how disciplines emerged at the University of 649). Similar developments are described by Gibbons and Göttingen in mid-eighteenth century in order to “facilitate colleagues in their argument for a “new mode 2” in the pro- ministerial paperwork.” Hence, the emergence of disciplines duction of knowledge. In this narrative disciplinary struc- as a prime unit for the organisation of knowledge is largely a tures and boundaries are seen as hindering knowledge pro- history of stabilisation, formalisation and bureaucratisation. duction: “Conformity is encouraged by disciplinary collegi- The development of disciplines was enabled when dis- ality, by expectations and rewards from disciplinary peers” tinct knowledge areas separated themselves from the three (2001, 149). According to Gibbons and colleagues, disci- faculties of the traditional university. Furthermore, the sta- plines are foremost articulated in teaching and education. bilisation of disciplines in the nineteenth century was de- While disciplinary identity matters greatly inside the univer- pendent on the development of “occupational roles.” The sity, it is less relevant when communicating with society at gradual processes of professionalisation and specialisation large. Such accounts suggest that disciplines are too narrow were instrumental in positioning “discipline” as the primary to address key problems in society, and disciplinary bound- unit of academic activity (Stichweh 1992). In an American ary keeping hinders the production of knowledge. The ideas context, disciplines emerged as important organisational of “mode 2 knowledge production” and “post-normal sci- and institutional entities even later, as Jacobs (2013, 45) ence” have had considerable impact on policies and research writes: “Before the Second World War, the idea of disci- agendas despite being criticised for being largely “political” pline-based departments was well ensconced. As a practical constructs rather than empirically grounded observations matter, however, at most colleges and universities, depart- (Godin 1998). In the wake of these criticisms, the discipli- ments were usually remarkable small in size.” nary structuring of research and education is questioned on During the twentieth century, disciplines came to be the several counts. Jacobs (2013, 13) summarises the five main main organising principle for the division of labour in aca- criticisms of disciplines, they: demic institutions, and while increasingly questioned they still play a central role in structuring and organising 1) hinder communication, knowledge. Hence, an important attribute of an academic 2) suppress innovation, discipline is the time horizon under which it functions, and 3) hamper economic contributions from universities, while research policy and institutional arrangements may 4) hinder the development of integrated solutions to change rather rapidly, disciplinary structures tend to remain urgent social problems and largely unchanged. At the same time, academic disciplines 5) result in the fragmented education of undergradu- in their present form are rather recent inventions, a little ates.

250 Knowl. Org. 47(2020)No.3 B. Hammarfelt. Discipline

4.3 The future of disciplines movement towards “postdisciplinarity” may be seen as an attempt to lessen the professional autonomy of academic re- Disciplines, it should be emphasised, do not play the same searchers. role across fields and contexts. In older traditional disciplines, for example philosophy or history, disciplinary iden- 5.0 Concepts for disciplinary boundary crossing tity is strong, despite recurring calls for interdisciplinarity. Moreover, as “disciplines” have strong connections to edu- Boundaries between disciplines are sometimes well de- cational programs, teaching and training, the concept seems fended, and disciplinary identity plays an important role in to be more central to academic fields in which teaching tra- many fields. Yet, due to intrinsic and extrinsic reasons ditionally has been important. This is reflected in the way “boundary crossing” has become pervasive in contemporary disciplines appear to play less important roles in fields academia, to the degree that “boundary crossing has become where a considerable part of research is conducted outside part of the process of knowledge production, not a periph- the university system and in fields with a “vocational” focus eral event.” (Klein 1996, 56). Klein identifies six partly over- (Klein 1996, 39). lapping reasons for crossing disciplinary borders: The importance of “discipline” is not least visible in the appointment of professors where a candidate may be ex- 1) changes in cognitive and epistemological structure, cluded, not on the basis of merits, but purely on the basis of 2) borrowing of tools, methods, theories and con- not having the “correct” disciplinary background. A disci- cepts, pline that demands proper training (e.g., PhD education) in 3) the pull of urgent social and intellectual problems, order to qualify as a professor is history, while the situation 4) the current complexity of disciplinary research, is different in a field such as economics where the ability to 5) relations with neighbouring disciplines and contribute to the field (e.g., publish in economic journals) 6) redefinitions of disciplinary borders. warrants inclusion in the discipline (Hammarfelt 2017). Thus, the borders surrounding disciplines are sometimes In addition, it might be added that changes in research policy, defended fiercely, while others are more permeable. Moreo- for example regarding funding opportunities, may be a fur- ver, when discussing “academic disciplines” it is important ther factor that encourages disciplinary boundary crossing. to emphasise that this organisation of knowledge produc- Interaction between disciplines takes many forms, and tion is more pronounced in continental Europe than in the depending on the level of depth and integration such activ- United Kingdom and the United States (Lawn and Furlong ities may be labelled as: “crossdisciplinary,” “multidiscipli- 2009; Gibbons et al. 2001, 149). nary,” “interdisciplinary,” “transdisciplinary” or even “post- Despite heavy criticism, disciplines remain important for disciplinary.” While these concepts overlap somewhat, they organising contemporary knowledge production and also carry specific connotations in terms of how boundary higher education. For example, while Gibbons et al. mainly crossing occurs. Just as in the “concept” of discipline, these view disciplines as out-dated and conservative constructs, concepts are rarely properly defined in the literature. How- they still contend that “disciplinary structures are long-term ever, a few attempts to differentiate between them have been and relatively stable.” Moreover, as Stichweh (1992) notes, made, and the following characterisation, which partially is researchers still believe in the intellectual rationality of a derived from Van den Besselaar and Heimeriks (2001), a few principal disciplinary identity. The importance of discipli- distinctive traits are outlined: nary identity is illustrated in that almost all “interdisciplinary” gatherings start with participants referring to their – “crossdisciplinarity” is any interaction between dis- “parent” discipline: origin still matters, even among highly ciplines, for example referencing literature from an- interdisciplinary researchers. other field, but it must not involve attempts of inte- For some, like Stichweh (1992, 14), disciplines are im- gration, portant as counterweight to rapid and extreme specialisa- – “multidisciplinary” research suggests that a subject tion. In fact, disciplines such as history and many other is studied using different disciplinary approaches, yet fields in the social sciences and humanities, which require there is little integration in terms of theory and find- certain broadness in both teaching and research, appear to ings, be most inclined to defend disciplinary borders. Moreover, – “interdisciplinarity” involves a further integration of disciplines are important in relation to professional auton- concepts, theory and methods, resulting in a more co- omy and collegiality, as these function as “reputational or- hesive and integrated approach to a certain problem, ganisations” (Whitley 2000), which act alongside and be- – “transdisciplinarity” is based on a shared theoretical yond institutional hierarchies and thus offer alternative ven- framework and methodology, which transcends re- ues for recognition and reward. In such interpretations, the search projects and problems.2

Knowl. Org. 47(2020)No.3 251 B. Hammarfelt. Discipline

Due to the high level of integration of both theories and pendent upon each other can, in fact, be questioned. The methods transdisciplinary fields may over time evolve into view of science as a type of “work organisation” might be new “disciplines.” This development is for example visible insufficient for explaining the distinctiveness of scientific in the field of gender studies. Finally, the concept of “post- reasoning across disciplines, but Whitley’s theory provides disciplinarity” is used less often in actual studies of science, an analytical position from which differences in communi- and is used more as an argument for an academic system cation structures between research fields can be studied. that completely has abandoned disciplinary borders, including departmental structures within universities. 6.2 Disciplinary tribes

6.0 Theories for understanding disciplinary Tony Becher and Martin Trowler in Academic Tribes and Ter- differences ritories: Intellectual Enquiry and the Culture of Disciplines (2001) offer an anthropological view of how disciplines are A comparative approach to studying academic disciplines is organised. They build upon differences between research common, where practices, patterns and structures in one fields that can partly be explained by categorisations such as field are related to another field. In the literature, we find “soft/hard” and “pure/applied” research. These characteris- two frequently used theoretical frameworks for explaining tics were identified previously by Biglan (1973), who in turn disciplinary differences; one developed by Whitley (2000) based his categorisation partly on the concept of “a paradigm” and the other by Becher and Trowler (2001) (Whitley’s (Kuhn 2012). According to this characterisation pure science framework focuses on organisational aspects while Becher is, in general, self-regulating whereas applied science is open and Trowler’s focus is anthropological thus giving slightly for influence from outside. different interpretations of the object under study. Becher and Trowler differentiate between “urban” and “rural” sciences and between sciences that are “convergent” 6.1. Social and intellectual structure as opposed to “divergent.” The distinction between rural and urban reflects how circumscribed a discipline or a re- A framework for studying disciplinary differences, or rather search area is; if many researchers are focused on the same differences between research fields, is introduced by Rich- problem, then the research area can be categorised as urban, ard Whitley, in The Intellectual and Social Organization of while the opposite is true for a rural one. Convergence indi- the Sciences (2000). The aim of the book is to contribute to cates the degree to which standards and procedures are an understanding of research fields “as particular kinds of agreed upon. A convergent discipline is guided by a control- work organizations which construct knowledges in differ- ling elite of researchers, while a researcher in a divergent dis- ent ways in different contexts” (6). Whitley’s theory is based cipline has greater freedom in choosing problems and meth- on two main axes: mutual dependency and task uncertainty. ods (Becher and Trowler 2001, 184-5). A related concept is Mutual dependency is a measure of how much the individ- Andersen’s (2016) idea of differences between disciplines in ual researcher is dependent on colleagues in his research. terms of “cognitive convergence,” which focus on similar as- The second axis in Whitley’s theory is the degree of task un- pects but from a more epistemological (and philosophical) certainty, which in turn depends on the intellectual organi- perspective. sation of a research field. The “Kuhnian view” would be Becher and Trowler (2001) propose an accessible theory that “the more paradigm bound a field is, the more predict- in which commonly used metaphors describe the character- able, visible, and replicable are research results and the more istics of research fields. Moreover, the concepts used by limited is permissible novelty” (119). In short, such a per- Becher and Trowler are drawn from different theories, and spective entails that a field with a high level of agreement on are, therefore, not part of a unified system, unlike Whitley’s overarching goals, theories and methodology is less likely to framework. The terms used to describe the different catego- develop path breaking discoveries. risations hard/soft, pure/applied and urban/rural are first of There is a strong focus on the organisation of work and all not specialised, and secondly, they are value laden. Usu- the practices of researchers in Whitley’s framework. The sig- ally something “pure” is regarded as better than the “ap- nificance given to these aspects partly explains its common plied,” and “urban” is connected to the modern while “ru- use in studies of differences in scholarly communication be- ral” could be associated with the past. tween fields (Talja et al. 2007) as well as in the development The choice of theory can be said to be dependent on how and institutionalisation of research fields (Åström 2004). “discipline” was defined in the first place; and the perspec- However, it can be argued that disciplines and research tive of “disciplines” as either social or epistemological enti- fields are more than work organisations, and the conclusion ties is of great importance. Notably, both these theories try that “the social organisation” and “the intellectual organisa- to combine elements relating both to the social structure tion” (or cognitive organisation) of research fields are de- and to intellectual and epistemological characteristics, and

252 Knowl. Org. 47(2020)No.3 B. Hammarfelt. Discipline some of their concepts, like divergence/mutual dependency, like structures, “research area” often targets smaller units. overlap considerably. While still very much used in studies Typically, it is used to refer to the research interests of indi- of academic fields it can be argued that Whitley’s and viduals or institutions (e.g., “scholarly communication” or Bechers and Trowler’s theories do not fully reflect contem- “urban living”), which often span across disciplinary bor- porary structures and developments in academic knowledge ders. production. Attempts have, therefore, been made to extend these frameworks in order to develop updated and more de- 7.3 Knowledge domain tailed accounts (cf. Gläser et al. 2018). Domain is used in LIS to represent a specific area of 7.0 Related concepts knowledge. The concept is associated with “domain analysis,” an approach introduced by Hjørland and Albrechtsen Discipline is often used in relation to similar concepts, such (1995), which emphasises the social and contextual nature of as research field, research area and knowledge domain. It is knowledge. Domain, in the definition provided by Hjørland not uncommon that these terms are used as synonyms in (2017) could take the form of a discipline, but it can “be dis- written text (Hammarfelt 2018), yet, as we will see below, tributed in multiple disciplines or specialties, or be a non-dis- these concepts all carry different meanings and connota- cipline, such as a hobby. Subject in this encyclopedia [ISKO] tions. Notably, topographical metaphors, such as field and is understood as the object of subject analysis, which is also a area, are often used to describe scientific disciplines and re- different concept. A domain, on the other hand, is a speciali- search specialties. Research directed at a specific topic is de- sation in the division of cognitive labor that is theoretically scribed as a “field” with “boundaries” to other neighbour- coherent or socially institutionalised.” The important dis- ing fields. Such metaphors also extend to concepts like “re- tinction is here made in terms of theoretical coherence, which search front” and “disciplinary landscape.” rarely is found in disciplines, expect perhaps in a few “paradigmatic” fields (such as physics) (for in-depth definition and 7.1 Research field / intellectual field conceptualisation see, Hjørland (2017) “domain analysis” http://www.isko.org/cyclo/domain_analysis). The intellectual field is an expansive and more general social unit of knowledge production and co-ordination. Whitley 7.4 Specialty (2000, 7) defines “intellectual field” as a “broader and more general social unit of knowledge production and co-ordina- In terms of size, specialty has much in common with “re- tion” compared to discipline. In several ways “intellectual search area” but it is often used in more specific terms. A fields” share many similarities with disciplines, as they are specialty is often regarded as a limited and coherent area of distinct and well-defined organisations that control how re- research. A definition given by Morris and Van der Veer search is done. Importantly, however, as Whitley points out, Martens (2008, 213) is that: “Research specialties consist of intellectual fields “are by no means always identical with relatively small self-organising groups of researchers that employment or educational unit boundaries.” Hence, “intend to study the same research topics, attend the same con- tellectual field,” can be regarded as a broader concept than ferences, publish in the same journals, and also read and cite discipline as it incorporates activities outside academia, alt- each others’ research papers.” De Solla Price estimated that hough both concepts share a connection to social and or- a scientific specialty or community comprises about 100 au- ganisational structure. The broader scope of intellectual thors or 10, 000 articles (Price 1963). Later calculations in- field, sometimes referred to more loosely as “research field,” crease these estimations to 250-600 authors (Wray 2010). may explain why the term is preferred when explicitly criti- Research specialties can be seen as “the largest homogenous cising traditional academic disciplines (c.f. Gibbons et al., unit in the self-organising system of science” that share 2001). Intellectual field or research field seems to be a viable knowledge, vocabulary and archival literature (Morris and alternative for describing units or groupings that have reso- Van der Meer Martens 2009, 219). In this view, research nance in the social organisation of research (for example, fields and disciplines are viewed as larger units where “local through reward structures) while reaching beyond the limi- homogeneities are mixed together and cannot be studied in tations of traditional disciplines. local terms” (219). Obviously, there are major differences across disciplines regarding the size of specialties, and par- 7.2 Research Area ticularly large-scale collaborations in physics, for example, may involve larger groups. Research area is a loosely defined term, which in its geo- Specialty is not the only term used to denote a group of graphical connotations relates to field. While the concept researchers focused on the same topic, method or theory; “research field” is quite often used to connote discipline- but a range of different concepts have been proposed, de-

Knowl. Org. 47(2020)No.3 253 B. Hammarfelt. Discipline pending on the focus of research. “Thought collectives” ferent definitions, conceptualisations and operationalisa- [Denkkollektiv], (Fleck [1935] 1979), “scientific communi- tions of the concept. ties” (Hagstrom 1965), “invisible colleges” (Crane 1972), In fact, Sugimoto and Weingart (2015) argue for an un- “scientific collectivities,” (Woolgar 1976), “scientific/intel- derstanding of disciplines as a boundary object in order lectual movements,” (Frickel and Gross 2005) and “epis- make room for “multiple interpretations.” However, the la- temic cultures” (Knorr Cetina 2009) are but a few of the bels we assign and the categorisations we create matter, not terms in use. only when presenting a “result” but also when formulating fundamental research questions. Excessive precision in de- 8.0 Discussion fining the concept of “discipline” is probably neither possible nor desirable, but there is undoubtedly a need for ex- The complexity of defining “discipline” is partly due to its plicit discussions and justifications when using it and re- dual role as both a social and organisational entity and as a lated concepts. Treating concepts such as “field,” “subject” description of specific areas of knowledge. This tension be- and “discipline” as synonymous results in a loss of concep- comes visible when studying the etymology of the concept, tual diversification and exactness, which may hamper theo- where the origin can be traced back to both the one who stud- retical and methodological development, and the imprecise ies (“disciple”) and what is studied (“doctrina”). These two use of concepts can hinder the transfer and incorporation of functions, as a way of categorizing and structuring knowledge research findings. Due to different understandings of what and as a principle of dividing work in academic institutions, constitutes a discipline, areas related to knowledge organisa- is evident also in theories developed for studying and compar- tion and LIS, such as the sociology of science, science and ing disciplines. Historical studies of disciplines might reveal technology studies (STS), studies in higher education and how emphasis on either the social or the epistemological and research policy may thus find it difficult to communicate intellectual has shifted over time. The organisational func- effectively and to incorporate insights from other fields. tion, which by many authors is seen as central to the develop- Moreover, vague use of “discipline” makes it difficult to ment of contemporary “academic discipline,” is tightly con- draw conclusions that are valid beyond a specific study. nected to the emergence of the modern university, while “dis- Thus, if the studied “disciplines” have little overlap with ac- cipline” in terms of ringing in a more generic “area of learn- ademic disciplines as they manifest themselves as organisa- ing” has a longer history. That our understanding of disci- tional units in contemporary academia, then the conclu- pline is context dependent is further emphasised by differ- sions drawn from such studies have limited use. For example ences in its use and significance across geographical regions, in the subfield of bibliometrics, the label “discipline” often where its importance is greater in continental Europe com- serves the function of hiding what is really studied, a very pared to the United Kingdom and the United States. Similar specific set of published documents in a database, and not a differences can also be found between disciplines where tradi- standalone organisational unit that is manifested in social tional, teaching-oriented fields are more prone to exhibit a arrangements (Hammarfelt 2018). strong disciplinary tradition. In conclusion, contemporary academic disciplines This is a conceptual paper and does not engage in longer should primarily be seen as social and organisation con- descriptions and the analysis of specific fields and disci- structs. The historical emergence of academic disciplines plines. However, the claim made is not that disciplines are supports such an understanding, as do theories on how for mostly conceptual constructs but rather that they are knowledge is produced and organised. Nevertheless, it is im- firmly grounded in research practices, organisational struc- portant to recognise and study how intellectual and episte- tures and materialities. Disciplines, in comparison with mological features interact with social structures, as disci- fixed constructs, are “enacted” in actions and activities plines are shaped and developed through the constant inter- (Pilerot 2014, 47). Indeed, an important function of disci- play of social organisation and intellectual formation. plines is training in a specific tradition of doing things; or in a Foucauldian interpretation: the disciplining of the body Notes to perform tasks in a certain manner. Ultimately, our perspective on disciplines is dependent on the purpose and the 1. Mills and Broughton (Bliss 1977, 37) argued strongly for perspective taken; for an information researcher, communi- the use of disciplines as the basic organizing principle in cation patterns and publication practices may be of key im- the introduction to the Bliss 2 system. The opposite portance, while a sociologist would focus on relations and viewpoint has been defended by, among others, James organisational features. Epistemological aspects emerge in a Duff Brown (cf., Beghtol 2004), Szostak (2008) and philosophical perspective, while the historian would natu- Gnoli (2016 and 2017). rally focus on the historical development of disciplines. 2. As pointed out by Van den Besselaar and Heimeriks Clearly, these different perspectives will result in slightly dif- (2001), an older interpretation of the term points to “in-

254 Knowl. Org. 47(2020)No.3 B. Hammarfelt. Discipline

terdisciplinarity” as a “meta-theoretical perspective” like Evans, Gillian R. 1980. Old Arts and New Theology: Begin- structuralism or Marxism. nings of Theology as an Academic Discipline. Oxford: Ox- ford University Press. References Fleck, Ludwik. (1935) 1979. Genesis and Development of a Scientific Fact. ed. Thaddeus J. Trenn and Robert K. Abbott, Andrew. D. 1988. The System of Professions: An Es- Merton, trans. Fred Bradely and Thaddeus J. Trenn. say on the Division of Expert Labor. Chicago: University Chicago: University of Chicago Press. Translation of En- of Chicago Press. stehung and Entwicklung einer wissenschaftlichen Tatsa- Andersen Hanne. 2016. “Collaboration, Interdisciplinarity che: Einführung in die Lehre vom Denkstil und and the Epistemology of Contemporary Science.” Stud- Denkkollektiv, 1935. ies in History and Philosophy of Science Part A 56: 1-10 Foucault, Michel. 1971. The Archaeology of Knowledge and Åström, Fredrik. 2004. “Library and Information Science the Discourse on Language, trans. Alan Sheridan. New in Context: The Development of Scientific Fields and York: Pantheon Books. Translation of L'Archeologie du their Relation to Professional Contexts.” In Aware and Savoir. Responsible: Papers of the Nordic-International Collo- Foucault, Michel. 1995. Discipline and Punish: The Birth of quium on Social and Cultural Awareness and Responsi- the Prison, trans. Alan Sheridan. New York: Vintage bility in Library, Information and Documentation Stud- Books. Translation of Surveiller et punir. ies, ed. Boyd. W. Rayward, (SCARLID). Lanham, MD: Frickel, Scott and Neil Gross. 2005. “A General Theory of Scarecrow, 1-27. Scientific/Intellectual Movements.” American Sociologi- Bawden, David. 2017. “Chemistry and its (Information) His- cal Review 70: 204-32. tory,” The Occasional Informationist: Irregular Thoughts Gibbons, Michael, Camille Limoges, Helga Nowotny, Si- on the Information Sciences (blog), September 3. mon Schwartzman, Peter Scott and Martin Trow. 2001. http://theoccasionalinformationist.com/blog/. The New Production of Knowledge: The Dynamics of Sci- Becher, Tony and Paul Trowler. 2001. Academic Tribes and ence and Research in Contemporary Societies. London: Territories: Intellectual Enquiry and the Culture of Disci- Sage. plines. Buckingham: Open University Press. Gieryn, Thomas. 1999. Cultural Boundaries of Science: Beghtol, Clare. 2004. “Exploring New Approaches to the Credibility on the Line. Chicago: University of Chicago Organization of Knowledge: The Subject Classification Press. of James Duff Brown.” Library Trends 52, no. 4: 702-18. Gläser, Jochen, Grit Laudel, Christopher Grieser and Uli Biglan, Anthony. 1973. “The Characteristics of Subject Meyer. 2018. “Scientific Fields as Epistemic Regimes: Matter.” Journal of Applied Psychology 57: 195-203. New Opportunities for Comparative Science Studies.” Bliss, Henry E. 1977. Bliss Bibliographic Classification. In- Working paper. Social Science Open Access Repository. troduction and Auxiliary Schedules. ed. Jack and Vanda https://nbnresolving.org/urn:nbn:de:0168-ssoar- Broughton. 2nd ed. London: Butterworths. 60196-2 Bourdieu, Pierre. 2003/1984. Homo Academicus, trans. Pe- Gnoli, Claudio. 2016. “Classifying Phenomena Part 1: Di- ter Collier. Cambridge: Polity Press. mensions.” Knowledge Organization 43: 403-15. Clark, William. 2006. Academic Charisma and the Origins Gnoli, Claudio. 2017. “Classifying Phenomena Part 2: of the Research University. Chicago: The University of Types and Levels.” Knowledge Organization 44: 37-54. Chicago Press. Godin, Benoît. 1998. “Writing Performative History: The Crane, Diana. 1972. Invisible colleges: Diffusion of New New Atlantis?” Social Studies of Science 28: 465-83. Knowledge in Scientific Communities. Chicago: Univer- Hagstrom, Warren. 1965. The Scientific Community. New sity of Chicago Press York: Basic Books Danziger, Kurt. 1997. Naming the Mind: How Psychology Hammarfelt, Björn. 2012. "Following the Footnotes: A Bib- Found its Language. London: SAGE. liometric Analysis of Citation Patterns in Literary Stud- Dewey, Melvil. 1979. Dewey Decimal Classification and ies." PhD diss., Uppsala Universitet. Relative Index. 19th ed. Albany, NY: Forest Press. Hammarfelt, Björn. 2017. “Recognition and Reward in the Dogan, Mattei. 2001. “Specialization and Recombination Academy: Valuing Publication Oeuvres in Biomedicine, of Specialties in the Social Sciences.” International Ency- Economics and History.” Aslib Journal of Information clopedia of Social and Behavioral Sciences. ed. Neil Management 69: 607-23. Smelser and Paul B. Baltes. Amsterdam: Elsevier, 14851- Hammarfelt, Björn. 2018. “What is a Discipline? The Con- 5. ceptualization of Research Areas and their Operational- Encyclopedia Mythica, s.v. “disciplina,” accessed May 5, ization in Bibliometric Research.” In 23rd International 2019. https://pantheon.org/articles/d/disciplina.html. Conference on Science and Technology Indicators (STI

Knowl. Org. 47(2020)No.3 255 B. Hammarfelt. Discipline

2018) September 12-14, 2018, ed. Rodrigo Costas, tät Zu Köln. Digitization of A dictionary, English and Thomas Franssen and Alfredo Yegros-Yegros. Leiden: Sanscrit. London: W.H. Allen and Co., 1851, s.v. “disci- Centre for Science and Technology Studies, 197-203. pline.” https://www.sanskrit-lexicon.uni-koeln.de/scans/ Hjørland, Birger 1998. “The Classification of Psychology: MWEScan/2020/web/webtc/indexcaller.php A Case Study in the Classification of a Knowledge Morris, Steven and Betsy van der Veer Martens. 2009. Field.” Knowledge Organization 25: 162-201. “Mapping Research Specialties.” Annual Review of In- Hjørland, Birger. 2016. “Knowledge Organization.” Know- formation Science and Technology 42: 213-95. ledge Organization 43: 475-84. Online Etymology Dictionary, s.v. “disciple,” accessed August Hjørland, Birger. 2017. “Domain Analysis.” Knowledge Or- 5, 2019. https://www.etymonline.com/word/disciple. ganization 44: 436-64. Oxford English Dictionary, s.v. “discipline,” accessed June 6, Hjørland, Birger and Hanne Albrechtsen. 1995. “Toward a 2019. New Horizon in Information Science: Domain Analy- Palmer, Carole and Melissa Cragin. 2008. “Scholarship and sis.” Journal of the American Society for Information Sci- Disciplinary Practices.” Annual Review of Information ence, 46: 400-25. Science and Technology 42: 163-212. Jacobs, Jerry. 2013. In Defense of Disciplines: Interdiscipli- Pilerot, Ola. 2014. “Design Researchers Information Shar- narity and Specialization in the Research University. ing: The Enactment of a Discipline.” PhD diss., Univer- Chicago: University of Chicago Press. sity of Borås. Kenny, Anthony J.P. and Anselm H. Amadio. 2019. “Aris- Price, Derek de Solla. 1963. Little Science, Big Science. New totle,” subsection “Lyceum.” In Encyclopedia Britan- York: Columbia University Press. nica. https://www.britannica.com/biography/Aristotle/ Ravetz, Jerome. 1999. “What is Post-Normal Science.” Fu- TheLyceum tures-the Journal of Forecasting Planning and Policy 31: Kimball, Bruce A. 2010. The Liberal Arts Tradition: A Doc- 647-54. umentary History. Lanham, MD: University Press of Simon, Josep 2011. Communicating Physics: The Produc- America. tion, Circulation and Appropriation of Ganot’s Textbooks Klein, Julie Thompson. 1996. Crossing Boundaries. in France and England, 1851-1887. Pittsburgh: Univer- Knowledge, Disciplinarities and Interdisciplinarities. sity of Pittsburgh Press. Charlottesville: University Press of Virginia. Stichweh, Rudolf. 1992. “The Sociology of Scientific Dis- Kluge, Friedrich. (1891). 2019. An Etymological Dictionary ciplines: On the Genesis and Stability of the Disciplinary of the German Language, trans. John Francis Davis. Structure of Modern Science.” Science in Context 5: 3-15 https://archive.org/details/etymologicaldict00kluguoft/ Stichweh, Rudolf. 2001. “Scientific Disciplines, History page/n5, accessed Jan. 28. Digitized version of London: of.” In International Encyclopedia of the Social and Be- George Bell, 1891. havioral Sciences, ed. Neil J. Semelser and Paul B. Baltes. Knorr Cetina, Karin. 2009. Epistemic Cultures: How the Sci- Oxford: Elsevier, 20: 13727-73. ences Make Knowledge. Cambridge, MA: Harvard Uni- Sugimoto, Cassidy and Scott Weingart. 2015. “The Kalei- versity Press. doscope of Disciplinarity.” Journal of Documentation 71: Kohler, Robert. 1982. From Medical Chemistry to Biochem- 775-94. istry. New York: Cambridge University Press. Szostak, Rick. 2004. Classifying Science: Phenomena, Data, Krishnan, Armin. 2009. “What are Academic Disciplines.” Theory, Method, Practice. Dordrecht: Springer. http://eprints.ncrm.ac.uk/783/1/what_are_academic Talja, Sanna, Pertti Vakkari, Jenny Fry and Paul Wouters. _disciplines.pdf 2007. "The Impact of Research Cultures on the Use of Kuhn, Thomas. 2012. The Structure of Scientific Revolu- Digital Library Resources.” Journal of the American So- tions. 4th ed. Chicago: University of Chicago Press. ciety for Information Science and Technology 58: 1674-85. Lawn, Martin and John Furlong. 2009. “The Disciplines of Tengström, Emin. 1993. “Biblioteks- och Informationsvet- Education in the UK: Between the Ghost and the enskapen: Ett Fler- eller Tvärvetenskapligt Område?” Shadow.” Oxford Review of Education 35: 541-52. Svensk Biblioteksforskning 1: 9-20. Lenoir, Timothy. 1997. Instituting Science: The Cultural Toulmin, Stephen. 1977. Human Understanding: The Col- Production of Scientific Disciplines. Stanford: Stanford lective Use and Evolution of Human Concepts. Princeton: University Press. Princeton University Press. Marrou, Henri I. 1934. “'Doctrina' et 'disciplina' dans la Turner, Stephen. 2000. “What are Disciplines? And How is langue des peres de l'eglise.” Archivum Latinitatis Medii Interdisciplinarity Different.” In Practising Interdiscipli- Aevi 9: 1-23. narity, ed. Peter Weingart and Nico Stehr. Toronto: Uni- Monier-Williams, Monier. (1851). 2014. Monier-Williams versity of Toronto Press, 46-65. English-Sanskrit Dictionary. XML ed. [Köln]: Universi-

256 Knowl. Org. 47(2020)No.3 B. Hammarfelt. Discipline

Vaan, Michiel de. 2008. Etymological Dictionary of Latin Woolgar, Steve. 1976. “The Identification and Definition and the Other Italic Languages. Leiden: Brill. of Scientific Collectivities.” In Perspectives on the Emer- Van den Besselaar, Peter and Gaston Heimeriks. 2001. “Dis- gence of Scientific Disciplines, ed. Gerard Lemaine, Roy ciplinary, Multidisciplinary, Interdisciplinary: Concepts Macleod, Michael Mulkay and Peter Weingart. Chicago: and Indicators.” In Proceedings of the 8th International Aldine, 233-45. Conference on Scientometrics and Informetrics, ed. Mari Wordreference.com, s.v. “discipline,” accessed May 30, 2019. Davis and Concepción Wilson. Sydney, Australia: Uni- https://www.wordreference.com/engr/discipline versity of New South Wales, 705-16. Wray, Brad. K. 2010. “Rethinking the Size of Scientific Dis- Whitley, Richard. 2000. The Intellectual and Social Organ- ciplines: Correcting Price’s Estimate.” Scientometrics 83: ization of the Sciences. Oxford: Oxford University Press. 471-76.

Knowl. Org. 47(2020)No.3 257 A. Smith. Physics Subject Headings (PhySH)

Physics Subject Headings (PhySH)†

Arthur Smith

American Physical Society, 1 Research Rd, Ridge, NY 11961,

Arthur Smith is Lead Data Analyst at the American Physical Society. He has worked in information technology roles for the Physical Review journals since 1995. He received a PhD in physics from Cornell University in 1991. He has given talks at Taxonomy Boot Camp (associated with the KM World conference), at the PIDapalooza meet- ings on persistent identifiers, and at a variety of related conferences.

Smith, Arthur. 2020. “Physics Subject Headings (PhySH).” Knowledge Organization 47(3): 257-266. 14 references. DOI:10.5771/0943-7444-2020-3-257.

Abstract: PhySH (Physics Subject Headings) was developed by the American Physical Society and first used in 2016 as a faceted hierarchical controlled vocabulary for physics, with some basic terms from related fields. It was developed mainly for the purpose of associating subjects with papers submitted to and published in the Physical Review family of journals. The scheme is organized at the top level with a two-dimensional classification, with one dimension (labeled “disciplines”) representing professional divisions within physics, and the other dimension (labeled “facets”) providing a conceptual partitioning of terms. PhySH was preceded in use by PACS (“Physics and Astronomy Classification Scheme”), which was in turn preceded by more ad hoc approaches, and this history and related vocabularies or categorizations will also be briefly discussed.

Received: 9 September 2019; Revised: 10 October 2019; Accepted: 31 October 2019

Keywords: physics, Physics Subject Headings (PhySH), concepts, classification, Phsyics and Astronomy Classification Scheme (PACS)

† Derived from the article of similar title in the ISKO Encyclopedia of Knowledge Organization, version 1.0, published 2019-07-31 Article category: KOS, specific (domain specific).

1.0 Introduction sification. These all allow for somewhat detailed subdivisions of the field, but (with the exception of Bliss, discussed later) The study of the natural world has been going on since al- their main subcategories for physics tend to focus on areas most the earliest recorded history, but “physics” as a distinct that were of interest 100 years ago, which have generally not field, as with most of the other major specializations in sci- been major research fields for many decades. Mechanics, for ence, dates to around the turn of the nineteenth century. example, is a relatively minor area of current physics research, The entities studied by physicists range from the simple to but occupies sections 531 to 534 in the UDC. Condensed the complex and from real-world objects to highly abstract matter physics, which accounts for close to half of current mathematical theories. Physicists try to understand physical basic physics research is confined to 538.9. Elementary parti- systems through detailed observations, reductionist analy- cle or high energy physics is not even mentioned in the UDC sis, and mathematical modeling, with a general (though far summary table (http://www.udcsummary.info); it is buried from uniform) division of labor between those doing exper- deeper in 539.1 which has the label “Nuclear physics. Atomic iments with real systems and those working on theories to physics. Molecular physics.” These classifications are suffi- explain or predict observational phenomena. cient for the relatively small number of books published in As a significant area of scientific research, physics has re- the field, but are not nearly comprehensive enough to use- ceived some attention in general knowledge organization or fully group the millions of scientific papers. Instead, much subject classification schemes: section 530 of the Dewey Dec- more detailed systems for knowledge organization have been imal, Melvil Decimal, and Universal Decimal Classifications developed by physicists themselves and particularly by the or- (UDC), subclass QC in the Library of Congress Classifica- ganizations publishing and providing indexes to physics re- tion, class C in Ranganathan’s Colon Classification (CC) search in recent decades. (Satija 2017), and class B in the Bliss Bibliographic Clas- 258 Knowl. Org. 47(2020)No.3 A. Smith. Physics Subject Headings (PhySH)

This article describes the development and structure of was “nuclear fission” (under “54—nuclear reactions and PhySH (“Physics Subject Headings”), a recently developed scattering” and “50—physics of nuclei”), etc. There was also controlled vocabulary for physics from the American Phys- a special classification code above the top subject-related ical Society. It also covers some other classification systems code—70—for errata (articles published as corrections to from related fields or previously used in physics. previously published articles). The decimal format resembled and may have been inspired by the Mathematics Sub- 2.0 History of knowledge organization in physics ject Classification codes which had used a two-digit + one or two-digit + two decimal notation starting in 1940 2.1 Physics classification in the nineteenth and early (https://mathscinet.ams.org/mathscinet/help/field_help. twentieth centuries html#mscp). This subject notation system soon had a profound effect As with much of the science literature, publications in physics directly on the structure of the journal. The 3,377 articles have grown exponentially since at least the 1950s. Forman et published in 1970 were split among four new journals: Phys- al (1975) found the number of articles published annually in ical Review A which covered the topics under headings 0, 10, the field in the late 1800s was on the order of 2000, with Ger- 20 and 30 (general, atomic, molecular, optical and fluid phys- many the leading nation for physics research at the time. The ics and related topics), Physical Review B had the topics under size of the English-language literature in the field was less than 40 (solid state physics), Physical Review C had 50 (nuclear half that total. The new Physical Review journal begun at physics), and Physical Review D had 60 (particles and fields). Cornell University in 1893 published only sixty-three papers These divisions remain today, although in 1995 Physical Re- in the year 1900, several of which were book reviews. view A was further split, adding a new journal Physical Review Even with a much smaller literature size than at present, E to cover some of the general, interdisciplinary, and fluid abstracting and indexing was found to be useful: in English, physics topics. Physical Review Letters, a weekly publication Science Abstracts started in 1898 (IET 1998) and included a for short high-impact papers that had been started in 1958, subject index from the first year. By 1902 Science Abstracts also in 1970 started sorting articles according to this decimal had split into a part A (Physics) and B (Electrical Engineer- notation with general physics papers at the front and particle ing), and after 1941 part A was simply known as Physics Ab- physics papers at the back of each issue. stracts; the publications formed the basis for the computer- based INSPEC (Information Service in Physics, Electro- 2.3 Physics and Astronomy Classification Scheme technology and Control) service in 1967, still with us today. (PACS) The Physical Review also had its own end-of-volume indexes before 1900, but these were initially just alphabetized Since the problem of classification in physics was not lists of author names and words pulled from article titles. In unique to the Physical Review journals published by the 1923, the author index and an “Analytic Subject Index” American Physical Society, in the early 1970s several organ- were separated; the latter was a flat alphabetized list of less izations got together to develop a more uniform system, than 100 classifying terms across all of physics, with a list of which became the Physics and Astronomy Classification matching articles provided under each term. Scheme (PACS). The process was described in an editorial by Krumhansl and Trigg in April 1975 in Physical Review 2.2 Physical Review hierarchical index Letters:

By 1964 the Physical Review had grown to about the size of Several years ago, negotiations began between the the entire worldwide physics literature of the year 1900 with American Institute of Physics (AIP), the major pub- 1873 published articles. The end-of-volume alphabetic sub- lishers of physics literature in English, and the Institu- ject lists seemed no longer sufficient, and a new hierarchical tion of Electrical Engineers (IEE), publishers of the classification with decimal notation was introduced by the principal English-language abstract journal, Physics editors. The ordering and hierarchy imposed by the nota- Abstracts. Eventually other discussants were brought tion had a logic that physicists would appreciate, with in, and the whole business was brought under the ae- higher numbers corresponding generally to higher energy gis of the Abstracting Board of the International phenomena. At the bottom was 0 (for general physics), and Council of Scientific Unions (ICSU-AB) … this at the top 60 (for particles and fields). Individual terms had scheme has been accepted by ICSU-AB, the AIP, and whole number or one-digit-after-the-decimal codes, for ex- the IEE, as well as some non-English abstracting jour- ample “classical mechanics” was 2 (under “0 - general”), nals … Reflecting this change, the grouping of papers “magnetohydrodynamics” was 33.4 (under “33—plasma in Physical Review Letters will also be changed to in- physics” which was part of “30—physics of fluids”), 54.9 corporate the PACS headings. Knowl. Org. 47(2020)No.3 259 A. Smith. Physics Subject Headings (PhySH)

Some examples of the PACS codes included “52.30.+r PACS outwardly resembled the Physical Review Hierarchical Plasma flow; magnetohydrodynamics” (under “52. The Index with hierarchy organized using a decimal notation that physics of plasmas and electric discharges,” and “50. Fluids, at the top level went from 00 to 99, with the ten top headings Plasmas and Electric discharges”); “24.80.+y Fission” (un- (00, 10, 20, etc.) representing the major subject areas. How- der “24. Nuclear reactions and scattering, general,” and “20. ever, it was internally significantly more detailed, having two Nuclear physics”), and “72.10.Jp Thermoelectric effects” digits after the first “.”, and then another “.” character fol- (under “72.10.-d Electronic conduction in metals and allowed by a final “+” (indicating no further hierarchy) or “-” loys,” “72. Electronic transport in condensed matter,” and (indicating there were additional terms at a lower level), or an “70. Condensed matter: electronic structure, electric, mag- upper-case letter for those lower-level terms. All of this was netic, and optical properties”). followed by a final lower-case letter used as a checksum (to PACS changed substantially over time. There were at least limit errors in manual retyping of the codes). In the final few twenty-three distinct versions from 1975 to 2010, growing editions of PACS, an additional layer in the hierarchy was from about 1,600 to almost 5,300 categories (some were also added by abandoning that final checksum letter, using a ‘-’ deleted on the way). The most dramatic year-to-year change character to indicate there were sublevels, and then the lower- was from 1976 to 1977, when there was a slight reordering level codes were added with a lower-case version of the first and about 30% more terms were added at once. letter, and a final lower-case letter that was simply incre- As a classification scheme, PACS was designed to allow a mented to indicate different sub-classifications. single code to characterize the full scope of a work, so the “leaf PACS also largely reversed the orientation of the earlier nodes” or narrowest classes of the scheme tended to combine index, ordering roughly by the distance scale involved rather multiple aspects or facets of the subject at hand, leading to than by energy, so that “10” now represented the highest en- some combinatorial complexity. As a result, the labels were ergy “elementary particles and fields,” “20” was now “nu- often not unique, with meaning determined not just by the clear physics,” while the burgeoning field of condensed label attached to the code in the scheme but to the hierarchy matter physics took both the “60” and “70” sections. As in which it sat. For example, the 1975 scheme had a second suggested by Krumhansl and Trigg in their July 1975 edito- entry with exactly the same “thermoelectric effects” label with rial, this necessitated a substantial reordering of the Physical code “72.20.Pa.” which placed it under “72.20.-i Conductiv- Review Letters table of contents. Those top-level groupings ity phenomena in semiconductors and insulators”—i.e., the for physics defined by PACS have had a much broader im- same phenomenon but in a different sort of system from the pact as they have been widely used in other classification metals and alloys of the other code. PACS was not consistent and related analysis work on physics, even very recently. For in whether it was the system or the phenomenon or some example, Desale and Kumbhar (2017) base their “first order other aspect that provided the finest layer of detail within a array divisions” (Table 3.4) largely on those top-level PACS class; in contrast to the thermoelectric example we can look at codes. Radicchi and Castellano (2011) in their analysis of the case of fullerenes, first introduced into the scheme with citation patterns in physics similarly subdivide the analysis the 1993 edition, and which appeared in ten different classi- according to that top-level PACS code. fications by the 2010 edition—see Table 1.

Code Label Parent 61.48.-c Structure of fullerenes and related hollow and 61. Structure of solids and liquids… planar molecular structures 68.35.bp Fullerenes 68.35.B- Structure of clean surfaces 68.55.ap Fullerenes 68.55.A- Nucleation and growth 71.20.Tx Fullerenes and related materials; intercalation 71.20.-b Electron density of states and band structure of crystalline solids compounds 72.80.Rj Fullerenes and related materials 72.80.-r Conductivity of specific materials 73.61.Wp Fullerenes and related materials 73.61.-r Electrical properties of specific thin films 78.30.Na Fullerenes and related materials 78.30.-j Infrared and Raman spectra 78.40.Ri Fullerenes and related materials 78.40.-q Absorption and reflection spectra: visible and ultraviolet 78.66.Tr Fullerenes and related materials 78.66.-w Optical properties of specific thin films 81.05.ub Fullerenes and related materials 81.05.U- Carbon/carbon-based materials (part of 81.05.-t Specific materials: fabrication, treatment, testing, and analysis)

Table 1. Fullerenes in the 2010 PACS edition. 260 Knowl. Org. 47(2020)No.3 A. Smith. Physics Subject Headings (PhySH)

This sort of complexity made PACS difficult for ordi- the organizations that helped develop PACS, but soon di- nary physicists to use; few practicing researchers could re- verged with what is now their proprietary Inspec classifica- member even a few of the codes relevant to their research. tion (IET 1998), which combines a modified version of the Full-text search and other electronic capabilities by the early PACS codes with additional keywords, providing a con- 2000s had made some of the other purposes of the classifi- trolled vocabulary that their indexers could use to help re- cation no longer relevant, so the owners of the classification trieval of relevant literature. AIP replaced PACS for their at the American Institute of Physics (AIP) decided to make own purposes with an internally developed thesaurus with 2010 the last edition of PACS. the primary goal of automated classification. This meant a rules-based system to look at abstracts or full text of articles 2.4 Other related vocabularies and classification and pick out terms or concepts that were likely to apply. schemes In closely related fields to physics, while there is conceptual overlap, the focus is often very different. In astronomy The BlissBibliographic Classification, one of the general (which was nominally covered by PACS in the “90” section) knowledge classification schemes mentioned in the intro- there are a number of different controlled vocabularies duction (and widely used by UK libraries), has been under- along different conceptual dimensions, exemplified by the going a major revision since 1969 (Mills and Broughton faceted search system of the Astrophysics Data System 1977) with the new version denoted BC2. Both the original (https://ui.adsabs.harvard.edu); one aspect of that is the (BC1) and new classifications have a prominent physics identifiers for specific celestial objects (a planet, star, gal- class with the physics schedule for BC2 published in 1999 axy), but specific objects or locations are not usually im- (Bliss Classification Association 1999). The Bliss classifica- portant to physics outside of astrophysics and geophysics. tion interestingly uses the same terminology of “disciplines” The Mathematics Subject Classification (MSC) codes and “facets” that was (apparently independently) adopted have been previously mentioned; these have always included by PhySH, though in BC2 the disciplines are only the high- many concepts from physics and some physics journals his- est-level classes (such as “physics” itself), and not narrower torically used these codes rather than PACS (which they re- subject areas. BC2 facets are organized with a technique of semble). Examples of physics terms from MSC are 81T55 “inversion” and “retroactive compounding” so that more (“casimir effect”), 76W05 (“magnetohydrodynamics and general groupings appear first, and there are many detailed electrohydrodynamics”), or 82D55 (“superconductors”). ordering rules. The notation is strictly ordered but does not Chemists have established a number of different taxono- reflect the hierarchy, which is noted separately. BC2 has nine mies of molecules and compounds and other chemical sys- main facets in its “standard citation order,” of which the tems, including the International Union of Pure and Ap- main ones used in physics are “operations & agents of oper- plied Chemistry International Chemical Identifier (In- ations” (methodology and techniques), “processes & prop- CHI) and the Chemical Entities of Biological Interest” erties,” “parts,” and “types;” for example under particle (ChEBI). In the life sciences there has been a proliferation physics (“BM”) there are elementary particle types listed of persistent identifiers for concepts, for example with the such as leptons (“BNM”) and quarks (“BNR”). However, Identifiers.org database (https://registry.identifiers.org/reg- many topics of current interest in physics research are not to istry) which allows a particular protein, gene, or other item be found in the BC2 physics class: examples are “quantum of interest to be identified with a “prefix: identifier” nota- information,” “nonlinear systems,” “optical trapping,” tion, enabling simpler indexing and searching. The Na- “graphene,” and “nanoscopics” or “mesoscopics” generally. tional Library of Medicine’s “Medical Subject Headings” “Fullerenes” are not listed under physics but can be found (https://www.nlm.nih.gov/mesh/) is a widely used curated in the chemistry (“C”) section with the rather lengthy nota- vocabulary of biomedical terms. tion “CGF LMG JQU.” Similar to PACS, the purpose of BC2 as a classification 2.5 Development of PhySH where items can be placed in a logical linear order results in the same subdivisions appearing among a variety of higher- PhySH had its origins in a November 2011 workshop in level topics. With BC2 the “filing order” does appear to be Boston (https://sites.google.com/site/physicsclassification more consistent and useful than the one chosen by PACS, 2011/) where the participants discussed what should follow but it leads to a similar combinatorial complexity of the PACS for classification and knowledge organization in scheme. The full BC2 physics schedule (Bliss Classification physics, as described by Smith (2019). Two basic models Association 1999) comes to eighty-nine pages with 5,231 stood out: centrally organized and comprehensive classifi- classifications. cation systems such as what the National Library of Medi- The Institute of Electrical Engineers (IEE, now the “In- cine did in the life sciences, or more independent and institute of Engineering and Technology,” IET) was one of teroperable “vocabularies,” often relying on the Simple Knowl. Org. 47(2020)No.3 261 A. Smith. Physics Subject Headings (PhySH)

Knowledge Organization System (SKOS) (Miles and Bech- tended to include several slightly different specializations. hofer 2009) design where each concept has a unique Uni- The URIs for the concepts in PhySH are actually Digital form Resource Identifier (URI). There was also much dis- Object Identifiers of the form: cussion of hierarchy (Smith 2019); the organization of parent-child and other relations depends in large part on the https://doi.org/10.29172/ purpose for which the vocabulary is being created, so good clear conceptual terms should be a first priority, with hier- where is an otherwise meaningless string consisting of archy a secondary component. the letters “a”-”f,” digits, and the “-” character. As a specific The purpose that the Physical Review journals had in example, “fullerenes” has the URI: mind was somewhat different than the concerns of other parties who had used PACS, according to Smith (2019). As- https://doi.org/10.29172/b755f66bb30d4ec1a7a12 signing submitted articles to editors with the right expertise, e31e5f675cb grouping related articles together when published, and as- sisting editors in finding referees were viewed as more im- The hierarchy within the realm of concepts is indicated us- portant than indexing automation or improving search ca- ing the standard “skos:broader” and “skos:narrower” rela- pabilities for end-users. So, the new scheme needed to be rel- tions, with “skos:related” also used for terms that are related atively easy for authors to assign to their papers from the but not in a parent-child manner. Note that the SKOS start, and for editors to check and correct if needed. In part, model allows arbitrary depth of hierarchy and multiple par- that meant the new scheme needed to be openly available to ents for any concept; rather than being a simple tree, the hi- anyone (not proprietary); it was also hoped that the scheme erarchy is described as a directed acyclic graph, with a partial would be easier to use and simpler (ideally smaller) than ordering from the broadest to the narrowest concepts. A PACS. concept with multiple parents is viewed as having a single Work on the new vocabulary began in earnest at the meaning—it does not matter through what path the con- American Physical Society in late 2013. The idea of group- cept was located, it should mean the same thing. In general, ing the terms with a faceted structure (such as “physical sys- PhySH concept labels are intended to be clear, unambigu- tems” or “techniques”) and filtering based on major re- ous, and independent of their parent or sibling terms. search areas of physics (such as “nuclear physics” or “con- There are a number of custom RDF predicates that densed matter”) allowed a reasonable partitioning of the vo- PhySH uses to indicate the special relations between con- cabulary so that different groups, including editors and out- cepts and the “disciplines” and “facets” they are assigned to. side consultants, could work on pieces relatively inde- There is also a special predicate used to identify deprecated pendently. The work was substantially complete by the first concepts. These are concepts that may be duplicative or oth- half of 2015, at which point serious internal testing began, erwise considered no longer relevant and are expected to be and after some more feedback the APS started using the new deleted in a future update. Deprecated concepts are not in- system in late 2015 (with initially only editors tagging pa- cluded in the statistics provided here. pers). There was then an “unveiling” in January 2016 (Conover 2016) and the Physical Review journals started re- 3.2 PhySH disciplines quiring authors to supply PhySH terms in the first half of that year. Shortly after that point, PhySH had completely The seventeen PhySH disciplines (see Table 2) are also identi- replaced PACS in handling of submitted manuscripts. fied by the same kind of URIs with preferred labels, but within the SKOS framework each discipline is treated as a 3.0 PhySH structure and usage “concept scheme” rather than a regular “concept.” Most of the disciplines have a relatively manageable number of con- 3.1 PhySH Concepts cepts of at most a few hundred, but several are quite large. Condensed matter physics has historically been the largest co- PhySH uses an adapted version of the SKOS model (Miles herent subfield, and as noted earlier was covered by two top- and Bechhofer 2009) for controlled vocabularies and the- level PACS codes from the start of the PACS scheme in 1975. sauri, so that every one of the 3,079 assignable terms in Some of the disciplines correspond to the other top-level PhySH is a “concept” with a unique preferred label and a PACS codes (“10” in PACS is “particles & fields” in PhySH stable identifying URI. Having a URI identifier allows the for example), but about half are new to PhySH: “accelerators label to be modified without any worry about losing assign- & Beams” and “physics education research” correspond to ments or relationships. Concepts may also have alternate la- relatively new specialty journals in the Physical Review family, bels to allow easier lookup when there are several different while “quantum information” and some of the others repre- words that express the same thing or when a concept is in- sent distinctive areas of recent growth in research activity. 262 Knowl. Org. 47(2020)No.3 A. Smith. Physics Subject Headings (PhySH)

PhySH Discipline Number of concepts Accelerators & Beams 103 Atomic, Molecular & Optical 387 Biological Physics 672 Condensed Matter & Materials Physics 1139 Fluid Dynamics 145 General Physics 133 Gravitation, Cosmology & Astrophysics 104 Interdisciplinary Physics 61 Networks 185 Nonlinear Dynamics 52 Nuclear Physics 134 Particles & Fields 239 Physics Education Research 19 Plasma Physics 338 Polymers & Soft Matter 862 Quantum Information 43 Statistical Physics 549

Table 2. The PhySH disciplines, as of version 1.1.1—these counts only include concepts directly listed in the discipline and their narrower terms.

PhySH facet label Number of disciplines Number of concepts Physical Systems 13 846 Professional Topics 2 12 Properties 4 53 Research Areas 17 1622 Techniques 13 752 – Computational Techniques 9 52 – Experimental Techniques 11 381 – Theoretical & Computational Techniques 2 226 – Theoretical Techniques 10 222

Table 3. The PhySH facets.

The sum of the concept counts in Table 2 (5,165) is concepts within the “nuclear physics” concept scheme, while siderably more than the total number of concepts in this “fluid dynamics research areas” and “fluid dynamics tech- version of PhySH (3,079). This is because the disciplines do niques” are similar top concepts within “fluid dynamics.” not in themselves constitute an exclusive partitioning of the These pairs are not strictly part of PhySH itself, rather conceptsthere is considerable overlap. The physical systems PhySH uses custom RDF classes and predicates (separate studied in one discipline may be studied in others, and tech- from the SKOS definitions) to define the facets and their niques are even more widely shared. relationships to disciplines and concepts. The “techniques” facet has been divided into subfacets 3.3 PhySH facets for the major classes of techniques; in the earliest versions of PhySH there was a three-fold split into “computational,” The facets in PhySH (Table 3) are similarly identified by “experimental,” and “theoretical.” The fourth subfacet URIs and preferred labels. However, as common cross-cut- (“theoretical & computational”) was a partial merger cre- ting classifications across all the disciplines (SKOS concept ated after finding significant overlap between the “compu- schemes), they do not strictly fit within the SKOS frame- tational” and “theoretical” subfacets. Aside from these sub- work at all. A SKOS-compatible version of PhySH is pro- facets, at the top level the PhySH facets do almost com- vided in which discipline-facet pairs serve as the “top con- pletely partition the concepts, with the total (using the 752 cepts” for each discipline. For example, “nuclear physics re- for “techniques” as a whole, rather than the numbers from search areas” and “nuclear physics techniques” are top con- its subfacets) amounting to 3,285, relative to the 3,079 dis- Knowl. Org. 47(2020)No.3 263 A. Smith. Physics Subject Headings (PhySH) tinct concepts. In other words, there is less than 10% overlap ment using PhySH, to provide the same level of characteri- of concepts between the top-level facets in PhySH. zation of the research. But note that both “electrical proper- The subfacet organization is probably unnecessarily ties” and “thin films” have refinements—narrower terms complex. A better organization (perhaps to be considered that could be more applicable, such as “bilayer thin films.” for a future PhySH release) would replace the “techniques” The PhySH approach allows for more precision on what ef- facet and its subfacets with two top-level facets, one for “ex- fect is being studied, what techniques are being used, or perimental,” and one for the merged “theoretical & compu- what systems being considered—but it is at the cost of re- tational.” These two subfacets have no overlap in the cur- quiring multiple terms to characterize the research. rent version and seem to be sufficient to satisfy the require- Note also that it is accepted and expected for a concept ments of classification. at (almost) any level of the hierarchy to be used in assign- The facets in PhySH are in principle similar to the ments, whereas with classifications one usually only assigns “PMEST” facets of Ranganathan (Satija 2017) but in prac- codes from the lowest possible level. There are no concepts tice somewhat distinct and perhaps not as generalizable. De- labeled as “other xxx” in PhySH; in PACS there were many sale and Kumbhar (2017) in their chapter five explicitly use such as “61.43.Er Other amorphous solids,” catchall catego- Ranganathan’s approach and assign many physics concepts ries whose meaning changed over time (narrowing as more to the “P,” “M,” and “E” facets; these assignments partially subclasses were added to the higher “disordered solids” align with the PhySH facets, so that “E” seems to be closely class). With PhySH one just uses the parent term (in PhySH related to PhySH “techniques,” “M” to “properties,” and it is “amorphous materials”), which encompasses all the var- “P” involves both “research areas” and “physical systems.” ious types. This is similar to the facets of the Bliss classification, at least One notable difference between PACS and PhySH is in with the BC2 “operations & agents of operations” corre- their ability to define a sorting order for sibling terms (con- sponding to PhySH techniques. BC2’s “processes & prop- cepts with the same “broader” or parent concept). With an erties” overlap more with PhySH “research areas” (and also alphanumeric notation as used by PACS, the notation de- “properties”), and “parts” and “types” correspond more to termines the order, and so if some particular order made the “physical systems” facet in PhySH. So, these two distinct sense, it was possible to enforce it through the notation. For library-oriented approaches to a faceted classification of PhySH the underlying identifiers are intended to be mean- physics show considerable overlap, but also some divergence ingless strings, so the only natural order is that of alphabeti- from what was done with PhySH. Note that all of these ap- zation of the preferred labels for concepts. For the most part, proaches avoid use of anything resembling Ranganathan’s the order is not important, but in a few exceptions judicious “S” and “T” facets in physics as physical concepts are sup- selection of labels helps to establish a preferred order. One posed to be applicable universally in space and time. example is within the “physical Systems” facet, “0-dimensional systems,” “1-dimensional systems” etc. are naturally 3.4 Using PhySH ordered, but would not be so if the verbal instead of the numeric form were used as the starting word. PhySH with its labeled concepts should be easier for regular With SKOS there is no enforced limit to hierarchy depth physicists to understand and remember than the alphanu- or to the number of siblings at one level in the hierarchy. meric codes used by PACS. The disciplines, facets, and hier- PhySH has a few concepts with paths as deep as nine levels archy as a whole are also designed to make browsing fruitful, (starting from the disciplines as level one, the discipline- but textual searching is easy enough and more frequently facet pairs as level two, and so on). The vast majority of con- used in practice. However, it should be noted that more cepts are found at levels four and five (two or three below PhySH terms are typically needed to characterize a scientific the discipline-facet usually set for filtering). Some concepts article than was true for PACS. have twenty or more direct “children” but usually the count PACS was indeed a classification (Hjørland 2017). That is much less. is, in principle there was one best code for each indexed doc- PhySH includes a number of terms like “fullerenes” that ument. That was why different facets were combined, such refer to specific materials or other specific physical systems, as with the fullerene examples in Table 1. For example, but only those with a high degree of interest to physicists. It 73.61.Wp on electrical properties of thin-film materials con- was anticipated that more detailed identifiers for specific taining fullerenes and related materials combines the con- physical systems—for example, a comprehensive list of ele- cepts of “electrical properties,” “fullerenes” and “thin-film ments, isotopes, crystallographic structures, astronomical materials.” In PhySH, each of those distinct concepts has objects, etc. would be better handled by outside vocabular- their own separate entry: “electrical properties,” “fuller- ies built with those domains in mind. So far PhySH has not enes,” and “thin films.” All three (plus anything else appli- been combined with these other vocabularies in practice, cable) would need to be assigned to the associated docu- but the ability is there in principle and should be relatively 264 Knowl. Org. 47(2020)No.3 A. Smith. Physics Subject Headings (PhySH) straightforward with concepts identified by their unique elementary concepts from several different facets, or from URI’s as is the practice with SKOS. different arrays in BC2 terminology (logical divisions) within a single facet. This enables grouping of documents 4.0 Governance, updates, and impact on several different dimensions, but it is preferential to only one such possible grouping which the classification design- PhySH is publicly available from a primary website at ers had some reason to believe would be most helpful. physh.org and also in a variety of downloadable formats In the paper-based world, the problem of arbitrariness in through github (https://github.com/physh-org/PhySH/). grouping was handled through indexes which allow alter- There have been a number of changes to PhySH since its nate paths to find relevant documents. With electronic doc- first use in 2015-2016, including the first public release in uments there is no longer any need to maintain a linear se- 2018 (version 1.0); the latest version as of this writing is quence and any grouping can be created dynamically. In the 1.1.1. It remains owned and governed by the American BC2 introduction (Mills and Broughton 1977), there is a Physical Society, but it is available for any other person or relevant discussion in section 4.8: organization to use under the Creative Commons CC-0 1.0 license (CC0). Suggestions for improvement are welcome Virtually all problems of information indexing are and may be submitted through the github site as github “is- problems caused by compound classes. In conven- sues.” The APS also receives suggestions from its authors tional indexing, the linking (intersection, coordina- and editors through online forms linked to the manuscript tion) of elementary terms to form compound classes handling system. Hundreds of suggestions have been acted is done at the time of indexing. In coordinate index- upon, with many new terms added and over one hundred ing, this is done only after receipt of a request. When concepts in the latest version marked as “deprecated.” the question is received then the search is made for the Guidelines for suggestions are provided including when a particular combination of elementary terms making new concept may be added, types of changes allowed for ex- up the search prescription. So it may be said that in isting concepts, rules for concept labels, and the possibility conventional indexing coordination to form com- of structural changes. pound classes is done before (pre-) receipt of any par- An internal group within APS reviews these change re- ticular request whereas in coordinate indexing it is quests, generally on a discipline-by-discipline basis. In 2018, done only after (post-) receipt of a request. the condensed matter & materials and fluid dynamics disciplines were reviewed, and some additional minor changes The label “concept” for the elementary ingredients of a were made in a few of the other disciplines. SKOS vocabulary suggests that these should always be con- So far, the only significant publicly known user of sidered as elementary terms, to be combined through such PhySH has been the American Physical Society which de- a post-coordination process. PhySH is indeed largely developed it, and has been using it since 2016 to index all arti- signed this way so that the individual concepts are mostly cles published in the Physical Review journals, about 70,000 elementary terms and are intended to be combined dynam- so far. APS has previously provided the PACS indexing data ically in searches, but even so there are many apparently of published articles for use by researchers (for example Ra- compound terms in the vocabulary. Some of these, for ex- dicchi and Castellano (2011) used this in their analysis), and ample “atomic & molecular processes in external fields,” are it is expected similar research may be conducted with the justified as providing a logical grouping for a large number PhySH terms in future. of more specific elementary concepts such as “photoemis- sion” or the “stark effect.” “Photon & charged-lepton inter- 5.0 Evaluation and considerations for the future actions with hadrons” on the other hand has no narrower terms below it, and perhaps could have been better repre- The type of ordering imposed by PACS and the more gen- sented by post-coordinating the two or three elementary eral classification schemes such as UDC, BC2, or CC was concepts involved. Nevertheless, over fifty articles have been necessary when our knowledge was primarily to be found in published in the Physical Review journals and indexed with paper documents stored on library shelves. The general idea this PhySH concept, so it may be justified as a useful pre- was to file items on similar topics together, and to file more coordinated term. general items before more specific ones if possible. The This brings us to the issue of “literary warrant,” dis- complexities of assigning one linear position to a document cussed by Barité (2018). A term that specifies over fifty rel- that covered multiple topics is exemplified in the “filing or- evant documents deserves to be included in the vocabulary. der” and consequent “citation order” specifications for the The guidelines for contributions (American Physical Soci- Bliss classification. Many of the detailed classes in PACS or ety 2018) specify criteria relating to a literary warrant for BC2 are the result of combining two, three, or even more new terms: Knowl. Org. 47(2020)No.3 265 A. Smith. Physics Subject Headings (PhySH)

Please consider if the concept you’re about to suggest of any individual to perceive all of it, good indexing and clas- is really needed. If a closely related concept already ex- sification will continue to be important long into the fu- ists, would adding the new proposal as an alias suf- ture. fice? Adding a new alias to an existing concept is preferred over adding a new concept .… A new concept 6.0 Conclusions may be needed where a significant body of work (dozens of papers per year, say) is associated with it, and Indexing and classification has a long history in the field of not distinguished by any existing concept in PhySH. physics. For scientific articles it was largely ad hoc until the advent of PACS in the 1970s, as a classification scheme with Whether these criteria were applied in the creation of decimal/alphanumeric notation. PACS established a widely PhySH in the first place may be questioned. Over 3,000 of used standard for both classification and ordering of the the 3,079 concepts have been used so far on published arti- subfields of physics for several decades. However, the rigid- cles, leaving about 2.5% of PhySH concepts that have not ity and complexity of PACS was not a good fit for the online been used at all. Almost all of these unused concepts (some world of the early twenty-first century, and in the last few examples are “leaves,” “kinesiology,” and “atactic poly- years a new approach following the philosophy of the SKOS mers”) come from the “biological physics” or “polymers and knowledge organization system produced PhySH, a pub- soft matter” sections of PhySH. These terms may eventually licly available (CC-0) controlled vocabulary and thesaurus be justified by articles published in other journals in the for physics managed by the American Physical Society. field so that the reason for no articles is simply that the Phys- PhySH disciplines are like the traditional subdivisions of ical Review journals publish relatively few articles in these physics, expanded to include new fields that have developed areas. But at least for now the literary warrant of some of the in recent decades. PhySH facets provide a cross-cutting clas- terms is questionable. Note that another roughly 300 sification by conceptual type, similar to the facets of Raga- (about 10%) have been used less than five times, while about nathan’s Colon Classification. PhySH concepts are identi- half the remainder have been used fifty or more times. fied by unique URIs while their unambiguous labels aid The concentration of unused terms in two disciplines human work in indexing and searching. The vocabulary is also suggests examining other potential inconsistencies be- still developing and has been made openly available for use tween the disciplines in PhySH. The issue of diverging sub- and feedback. There are some obvious improvements still facets has already been mentioned (some disciplines have needed, particularly in the organization of the facets, but it separate “computational techniques” and “theoretical tech- can be used now for indexing and classification in physics. niques” groupings, while others use a combined subfacet). As the system is largely compatible with the SKOS model There is also quite inconsistent treatment of the “proper- for knowledge organization, it will be interesting to see in ties” facet, which is so far not used by many of the disci- future how PhySH and other SKOS vocabularies can work plines. For example, “symmetries” in the “properties” facet together in organizing research in the physical sciences. of the “particles & fields” discipline is closely related to “symmetries in condensed matter,” but that has been placed References under the “techniques” facet in “condensed matter & materials physics.” The “research areas” facet in that discipline American Physical Society. 2018. “PhySH contribution contains several top-level terms that actually use the word guidelines.” https://physh.org/contribute “properties:” “electrical properties,” “structural proper- Barité, Mario. 2018. “Literary warrant.” Knowledge Organ- ties,” etc. It is not clear whether this inconsistent placement ization 45: 517-36. is deliberate or simply a consequence of different people Bliss Classification Association. 1999. Bliss Bibliographic making different decisions about use of the facets. Perhaps Classification. 2nd ed. Class B Physics. Full Schedule. Avail- some of these issues will be improved upon in future releases able at: http://www.blissclassification.org.uk/ClassB/B of PhySH. _sched.pdf Smith (2019) discusses the potential impact of “artificial Conover, Emily. 2016. “New Physics Classification Scheme intelligence” on the need for classification and indexing sys- Unveiled.” APS News 25, no. 2: 3. tems like PhySH and counters that rather, the manual work Desale, Sanjay and Rajendra Kumbhar. 2017. Methodology of classification is an essential ingredient to successful auto- to Develop Depth Classification Scheme for Physics. Er- mation in scientific research. PhySH was specifically de- furt: Lambert Academic. signed to be easy for those with field expertise (the authors Forman, Paul, John L. Heilbron and Spencer Weart. 1975. and editors) rather than classification experts or automated “Physics circa 1900: Personnel, Funding, and Productiv- tools to assign useful indexing concepts to articles. With the ity of the Academic Establishments.” Historical Studies scientific literature continuing to grow beyond the capacity in the Physical Science 5: 1-185. 266 Knowl. Org. 47(2020)No.3 A. Smith. Physics Subject Headings (PhySH)

Hjørland, Birger. 2017. “Classification.” Knowledge Organ- Mills, Jack and Vanda Broughton. 1977. “Introduction.” ization 44: 97-128. Transcription from Bliss Bibliographic Classification. 2nd IET (The Institute of Engineering and Technology). 1998. ed. “Introduction and Auxiliary Schedules.” London: “History of Science Abstracts and Inspec.” https:// Butterworths, [1-107]. http://www.blissclassification. www.theiet.org/publishing/library-archives/the-iet-ar org.uk/Class1/introduction.pdf chives/iet-history/history-of-science-abstracts-and-in Radicchi, Filippo and Claudio Castellano. 2011. “Rescaling spec/ Citations of Publications in Physics.” Physical Review E Krumhansl, James A. and George L. Trigg. 1975. “Indexing 83: 046116. and Classification.” Physical Review Letters 34: 1065-6. Satija, Mohinder P. 2017. “Colon Classification (CC).” Miles, Alistair and Sean Bechhofer, eds. 2009.” SKOS Simple Knowledge Organization 44: 291-307. Knowledge Organization System Reference.” https:// Smith, Arthur. 2019. “From PACS to PhySH.” Nature Re- www.w3.org/TR/2009/REC-skos-reference-20090818/ views Physics 1: 8-11.

Knowl. Org. 47(2020)No.3 267 T. Midtgarden. Peirce’s Classification of the Sciences

Peirce’s Classification of the Sciences†

Torjus Midtgarden

The University of Bergen, Centre for the Study of the Sciences and the Humanities, PB 7805, 5020 Bergen, Norway,

Torjus Midtgarden (born 1962) is Professor at the Centre for the Study of the Sciences and the Humanities at the University of Bergen. His research interests are social and political philosophy, philosophy of science and American pragmatism, in particular the philosophy of Charles S. Peirce and John Dewey. He has contributed to several book projects and published articles on pragmatism and its relevance for contemporary philosophy in journals such as Transactions of the Charles S. Peirce Society, Journal of the History of Philosophy, Contemporary Pragmatism, Euro- pean Journal of Social Theory, Philosophy and Social Criticism, Semiotica and Revue Internationale de Philosophie.

Midtgarden, Torjus. 2020. “Peirce’s Classification of the Sciences.” Knowledge Organization 47(3): 267-278. 32 references. DOI:10.5771/0943-7444-2020-3-267.

Abstract: Charles Peirce’s classification of the sciences was designed shortly after the turn of the twentieth century. The classification has two main sources of inspiration: Comte’s science classification and Kant’s theoretical philosophy. Peirce’s classification, like that of Comte, is hierarchically organised in that the more general and abstract sciences provide principles for the less general and more concrete sciences. However, Peirce includes and assigns a superordinate role to philosophical disciplines which analyse and provide logical, methodological and ontological principles for the specialised sciences, and which are based on everyday life experience. Moreover, Peirce recognises two main branches of specialised empirical science: the natural sciences, on the one hand, and the social sciences, the humanities and psychology on the other. While both branches share logical and methodological principles, they are based on different ontological principles in studying physical nature and the human mind and its products, respectively. Peirce’s most basic philosophical discipline, phenomenology, transforms his early engagement with Kant. Peirce’s classification of aesthetics, ethics and logic as normative sub-disciplines of philosophy relate to his philosophical pragmatism. Yet his more overarching division between theoretical (philosophical and specialised) sciences and practical sciences may be seen as problematic. Taking Peirce’s historical account of scientific developments into consideration, however, I argue that his science classification and its emphasis on the interdependencies between the sciences could be seen as sustaining and supporting interdisciplinarity and interaction across fields of research, even across the divide between theoretical and practical sciences.

Received: 17 December 2019; Revised: 30 January 2019; Accepted 22 February 2020

Keywords: sciences, science classification, Charles Sanders Peirce

† Derived from the article of similar title in the ISKO Encyclopedia of Knowledge Organization, Version 1.0, published 2019-11-21, last edited 2019-11-26. Article category: KOS, specific (domain specific). The author is greatful to colleagues at The Centre for the Study of the Sci- ences and the Humanities at the University of Bergen for their useful comments. He especially wishes to thank Mimi E. Lam and Silje Aambø Langvatn as well as the two anonymous referees and the IEKO editor for their useful suggestions.

1.0 Biographical and intellectual context and fication of the sciences from shortly after the turn of the sources of inspiration twentieth century.4 Notably, his pragmatism is reflected in his efforts to understand science as “a pursuit of living men” Charles S. Peirce (1839–1914)1 was the founder of philo- (CP 1.232) and a “living historic entity” (CP 1.44), rather sophical pragmatism and a pioneer in the development of than as “a mere abstract definition” (CP 1.232) or as “sys- formal logic.2 He was educated both as a philosopher and as tematized knowledge on the shelves” (CP 1.234). One over- a natural scientist, and he was well-informed about develop- all aim of his classification was thus to take account of actual ments in several disciplines. Both his philosophical pragma- developments in the natural sciences, but also in the social tism, which is concerned with the connections between sciences, psychology and the humanities. Across fields of re- knowledge and action,3 and his familiarity with several fields search, he saw processes of specialisation as a defining trait of scientific research inform and guide his elaborate classi- of modern science. In concrete sociological terms, he iden- 268 Knowl. Org. 47(2020)No.3 T. Midtgarden. Peirce’s Classification of the Sciences tified the results of these processes of specialisation by con- sification in efforts to provide a philosophical and non-psy- sulting “the list of scientific periodicals and the list of scien- chologistic basis for formal logic.8 This classification serves tific societies” (CP 1.237). Besides his studies of scientific as a starting point for his far more comprehensive science developments, however, his classification of the sciences classification from around the turn of the century. Moreo- had philosophical sources of inspiration. The two perhaps ver, while the young Peirce endeavored to develop Kant’s most imporant sources were Auguste Comte’s influential transcendental analysis of the categories of the understand- classification of the sciences and Immanuel Kant’s theoret- ing (“Verstand”),9 his later work in mathematical logic leads ical philosophy. him to derive categories formally from mathematics, and As Chiara Ambrosio has pointed out (2016), Peirce’s then to assign to philosophy the task of exploring and spec- classification of the sciences was inspired and influenced by ifying the content of these categories through experience.10 several other science classifications developed during the As I expand below, these categories, which the later Peirce nineteenth century, such as those of Jeremy Bentham names “firstness,” “secondness” and “thirdness,” inform ([1817] 1983) and William Whewell (1847). Yet it was the several divisions of his comprehensive science classification. classification of the father of positivism, Auguste Comte Yet, through his later category analysis, he modifies his early (1855), which influenced Peirce the most. Peirce’s classifica- Kantian conception of philosophy. In his later view, philos- tion shares with that of Comte that it reflects historical de- ophy considers experiential phenomena in an everyday life velopments in both social and natural science. In fact, Peirce sense without using specialised methods or drawing on re- even notes particular developments in the humanities, such sults from specialised scientific inquiries. In Peirce’s own as his perception that “linguistics is becoming more and terms (EP2, 146), it “contents itself with observations such more nomological” (EP2, 39) or oriented toward discover- as come within the range of every man’s normal experience, ing general laws. Yet, like Comte, Peirce is concerned with and for the most part in every waking hour of his life.” In an the unity of the sciences. From Comte’s classification, era of incessant scientific specialisation, Peirce thus gives Peirce takes the idea of an order of unilateral dependency pride of place to phenomena of everyday life in philosophy, between the sciences: the more general and abstract sciences decades before Edmund Husserl turned phenomenology provide principles for the more particular and concrete sci- toward the pre-scientific life world (1970). In Peirce’s clas- ences.5 In particular, Peirce, too, takes mathematics to be the sification, philosophy uncovers a shared experiential basis most abstract science upon which other sciences depend. for all scientific specialisation and to which specialised re- Nevertheless, Peirce’s classification diverts from that of searchers may need to take recourse in clarifying and rein- Comte in several ways. Peirce does not see mathematicians terpreting scientific concepts and theories.11 as making assertions about facts but rather takes them to de- With Peirce’s sources of inspiration in mind, I now take velop hypotheses about purely imagined possibilities and to a closer look, firstly, at the elements and formal architecture draw consequences from the hypotheses.6 Moreover, among of his classification of the sciences. Secondly, I consider the the specialised empirical sciences Peirce introduces a general categories lying at the heart of several of the divisions of the division between the natural sciences, on the one hand, and classification. Thirdly, I discuss the current relevance of the the social sciences, psychology and the humanities, on the classification. other. He thus recognizes variability and diversity between scientific subject matters and objects of knowledge to an ex- 2.0 The structure and elements of Peirce’s tent not found in Comte’s classification. More saliently, classification however, while metaphysics or philosophy generally is absent from Comte’s classification, Peirce assigns to philoso- In his classification, Peirce introduces a overarching tripar- phy the status of the second most general science after math- tite division between three branches of science: 1) science of ematics. Drawing also on his own work in formal logic, he discovery, which includes the sciences mentioned above; 2) assigns a key role to philosophy as the provider of logical, science of review, which encompasses any science classifica- methodological and ontological principles for all specialised tion, as well as history of science (EP2: 258–9, 458); and, 3) empirical sciences. practical science or science “for the uses of life” (CP 1.239), The role of philosophy in Peirce’s classification should for example, “pedagogics, … vulgar arithmetic, horology, be seen against the background of his Kantian legacy and as surveying, navigation, … librarian’s work” (CP 1.243).12 Alt- inspired by Kant’s idea of the “architectonic” character of hough Peirce’s classification focuses mostly on sciences of building philosophical systems in the Critique of Pure Rea- the first branch, the fact that the two last branches are in- son (CP 1.176).7 Already the early Peirce develops a classifi- cluded may give pause to reflect on their significance for the cation of philosophical disciplines based on an epistemolog- classification as a whole. ical analysis strongly influenced by Kant’s first Critique. Through including science of review, Peirce may be seen Later, in the 1890s, he reinvigorates and develops this clas- to reflexively recognize that a science classification is itself a Knowl. Org. 47(2020)No.3 269 T. Midtgarden. Peirce’s Classification of the Sciences scientific activity of a certain kind. Yet, while a science clas- much as if it had transported our race to another planet” sification is empirically constrained by its objects—histori- (CP 5.513). Given that not only technological products and cally evolving disciplines—its more general distinctions and services, but also technological research infrastructures and divisions cannot be established simply by describing and la- scientific knowledge, started to leave laboratories and aca- belling given scientific journals or societies. Since all classi- demic institutions in young industrial America, a division fication, on Peirce’s account, “is the arrangement of objects between pure and practical science becomes increasingly according to ideas” (CP 1.231), a science classification difficult to draw. Another consideration to the same effect would need to develop general classificatory ideas by fram- comes from observing genealogical conditions of science. ing and testing conjectures, just like in any research process Peirce notes how sciences “have grown out of the useful in the sciences of discovery.13 This calls for further reflexivity arts” (CP 1.226). For example, he points out (EP2, 38-9) and caution in conceptualising general divisions of the sci- that “the steam engine made mechanical precision possible ence classification, such as the division involving science of and needful” and that “[m]echanical precision rendered review itself and the other two branches of science. As for modern observational precision possible, and developed it.” the need to develop classificatory ideas and to go beyond From this genealogical perspective, too, and given Peirce’s mere descriptions of existing scientific institutions, one commitment to account for science as a “living historic en- could with Ambrosio further see Peirce’s science classifica- tity” (CP 1.44), a demarcation line between purely scien- tion as “a response to external constraints and pressures di- tific and practical disciplines is not ready made and would rectly related to the social organization of the sciences in the need to be decided more carefully in each case. nineteenth century” (2016, 4 emphasis original). Like sev- As mentioned, the main thrust of Peirce’s classification eral of the authors to which Peirce refers, such as Bentham concerns further divisions within science of discovery. and Comte, Peirce could thus be taken to respond to chal- Again, the main division is tripartite: mathematics, philoso- lenges related to the institutional organisation of the sci- phy and specialised empirical sciences. ences, to a need for reform in higher education, as well as to While this tripartite division may not seem controversial, more bureaucratic needs to survey and index all existing sci- Peirce construes and conceptualises hierarchical relation- entific publications in an era of incessant specialisation and ships of dependence between the sciences, and he assigns internationalisation of science. For the internal divisions of distinct roles to mathematics and philosophy. Notably, Peirce’s science classification, however, this suggests that Peirce’s view that philosophy would need to draw on math- there would be interactions or interconnections between ematics while mathematics needs no foundation in logic sciences of review and practical sciences. (which for Peirce is a philosophical discipline), contrasts Peirce’s division between science of discovery and practi- sharply with the logicist views of Alfred North Whitehead cal science might also seem to be a bit too straight forward, and Bertrand Russell (1910-13) and the logicism of the Vi- given his more pragmatist qualifications of scientific activ- enna Circle of logical positivism.14 Moreover, the way in ity. For example, in reflecting on the extensive social and which Peirce assigns a certain foundational role to philoso- practical consequences of modern science, Peirce recognises phy is peculiarly his own. Through its various sub-divisions, how science has transformed our everyday world “with its he points out, philosophy is to provide logical, methodolog- microscopes and telescopes, with its chemistry and electric- ical and ontological principles for specialised empirical dis- ity, and with its entirely new appliances of life, … almost as ciplines. Yet, by drawing on everyday life experience, as well

Figure 1. Peirce’s classification of the sciences, 1903 (simplified hierarchical structure). 270 Knowl. Org. 47(2020)No.3 T. Midtgarden. Peirce’s Classification of the Sciences as on conceptual resources from mathematics, philosophy In Peirce’s classification, logic is not only conceived as a itself becomes an empirical and in principle fallible science. normative but as a semiotic science further divided into In fact, Peirce invents and calls his most basic philosophical three sub-disciplines. “Speculative grammar” is “the general discipline “phenomenology” (or “phaneroscopy”15), which theory of the nature and meaning of signs” (EP2, 260); “log- is to explore the three categories that we consider more ical critic” concerns the validity of arguments in the formal closely in the next section. Despite the name chosen, sense; and “methodeutic” or “speculative rhetoric” consid- Peirce’s idea of this discipline is not historically related to ers methodological principles for specialised scientific in- Edmund Husserl’s phenomenology. Still, Peirce’s stress on quiries but also suggests principles for science communica- the need to develop a distinct philosophical and experiential tion and scientific writing.19 Peirce’s semiotic account of inquiry, independent of specialised science and psychology logic clearly differs from a purely formal approach in insist- in particular,16 suggests parallels to Husserl. ing on the requirement of studying the nature of signs in Another novelty in Peirce’s classification comes with the general, and the everyday use of linguistic signs in particular, further sub-division of philosophy into normative science prior to developing and using a formal logical system.20 and metaphysics. Normative science consists of three philo- As his last philosophical science, in addition to phenom- sophical disciplines that are hierarchically ordered: aesthet- enology and the normative sciences, Peirce introduces met- ics, ethics and logic. This subdivision shows another prag- aphysics. In its present condition, he admits, metaphysics is matist aspect of his classification. He makes logic, as the sci- only “a puny, rickety, and scrofulous science” (EP2, 375). ence of “self-controlled, or deliberate, thought,” dependent Yet, there is need, he thinks, for a general metaphysics which on ethics, which concerns “self-controlled, or deliberate, “seeks to give an account of the universe of mind and mat- conduct” (EP2, 160). Ethics, which distinguishes right from ter” (EP2, 259). Moreover, intimately linked to this philo- wrong, must in turn “appeal to esthetics for aid in determin- sophical science are two subordinated “nomological” sci- ing the summum bonum” (EP2, 160) since aesthetics studies ences that specify ontological principles for two branches of the very “admirableness of an ideal” (EP2, 142). In constru- specialised science of discovery: on the one hand, the natural ing logical reasoning on the model of self-controlled moral sciences (or what Peirce generally calls “physical sciences”); conduct, Peirce takes the reasoner to be guided by an over- on the other hand, the social sciences, humanities and psy- arching ideal (the truth) and by general patterns or “norms” chology (or what he generally calls “psychical sciences”).21 of right reasoning that are applicable in particular cases (CP Adopting an Aristotelian distinction, Peirce’s nomological 1.606). Despite his pragmatist construal of reasoning, how- sciences distinguish “efficient causation” as the underlying ever, he distinguishes the normative sciences generally as principle of the natural sciences, and “final causation” as the theoretical sciences and qualifies ethics in particular as a the- principle of the social sciences, the humanities and psychol- oretical inquiry into “what the fitness of an ideal of conduct ogy.22 As for the classificaton of psychology in particular, we consists in” (CP 1.600 emphasis original). The normative may note that Peirce was greatly impressed by the new ex- sciences are thus distinguished from the practical sciences, perimental psychology of Gustav Fechner and Wilhelm which are not treated as “integrant parts” of the normative Wundt, and that he even conducted pioneering research in sciences as such (EP2, 198) but which are rather concerned this new field.23 Nevertheless, he takes these new develop- with the conformity of action to a “given particular ideal” ments as mainly introducing new methods and not as rede- (CP 1.573). Nevertheless, the inclusion of the normative fining psychology as a natural science, or a science studying sciences in the overall classification still suggests that they processes defined according to the principle of efficient could or even should be linked to the practical sciences. In causation.24 Moreover, within each of the two branches of fact, Peirce considers how practical sciences “may be proba- specialised science he further recognises two subordinated bly expected to receive aid from [the Normative Sciences]” types of sciences that apply the principles formulated by the (EP2, 198), and he comments in particular that the exclu- nomological sciences: classificatory and descriptive sciences. sion of ethics from the practical sciences of “diplomacy and An example of a descriptive natural science is geology, while economics is immense folly” (CP 1.251). These suggested in the humanities history serves as an example of descriptive links to practical life might recall Comte’s classification and science.25 Yet Peirce also suggests a more dynamic relation positivism generally, which is closely connected to ideas of between the hiearchical levels. For example, while in the hu- moral progress and social reform.17 Yet, since such links manities and the social sciences he takes linguistics and an- come through philosophy, not sociology, as in Comte’s clas- thropology to be classificatory sciences, he observes that sification, they would involve a more distanced and no di- “[l]inguistics is becoming more and more nomological rect commitment to social intervention. Notably, however, [and] [a]nthropology is tending that same way” (EP2, 39). Peirce does propose a certain extended and critical applica- The division between two branches of specialised science tion of his normative sciences to our inherited moral beliefs can be seen to bear a certain similarity to Wilhelm Dilthey’s and their applicability to a modern technological society.18 (1989) division between sciences based hermeneutically in Knowl. Org. 47(2020)No.3 271 T. Midtgarden. Peirce’s Classification of the Sciences

Figure 2. Peirce’s classification of the sciences, 1903 (simplified presentation).

principles of understanding (“Verstehen”) and sciences ries as “modes of being” (CP 1.23). As modes of being, the founded in principles of explaining according to general categories would constitute the ways all phenomena in the laws (“Erklären”). However, in so far as Peirce’s philosophi- world (including ourselves) are. Moreover, as for their foun- cal sciences lay down logical and methodological principles dational role for specialised empirical sciences and for meta- for all specialised sciences, his classification still differs sig- physics, the categories would be basic for any conceptualisa- nificantly from Dilthey’s dualistic scheme. Moreover, his tion of what scientific theories are about. More indirectly, classification cuts across Dilthey’s scheme in grouping hu- through defining what a “sign” is, and by informing Peirce’s manities disciplines under nomological sciences, and his ob- comprehensive classification of signs,27 the categories are preservation that “[li]nguistics is becoming more and more no- supposed by any logical and scientific thinking, in so far as “all mological” (EP2, 39) is to the same effect. In addition, his thinking is performed in Signs” (EP2, 447) or “is of the sym- phenomenology and his speculative grammar uncover com- bolic nature” (EP2, 307). Accordingly, his phenomenological mon experiential and semiotic ground between the two analysis of the categories forms the backbone of his specula- branches of specialised science. In so far, his philosophical tive grammar, which analyses the nature of signs and their approach may be seen to have more in common with Mar- meaning. Nevertheless, although Peirce’s phenomenology tin Heidegger’s (2008) and Hans-Georg Gadamer’s philo- draws on his mathematical logic of relations, his various out- sophical hermeneutics (2004) than with Dilthey’s herme- lines of a phenomenological analysis are rather sketchy and neutics. lack the methodological sophistication that we may find in Despite the dual branching of the specialised empirical his semiotic and logical writings. Let us first consider how he sciences, we have noted several triapartite divisions in his distinguishes his three categories. classification. These tripartite divisions suggest the need to Firstness, in its most general and primary sense, is “what is consider his three categories more closely and the role of positively there,” “regardless of what is absent,” past and pre- phenomenology in the overall classification. sent (EP2, 150). Further, firstness is what is “sole and unique,” involving no reference to, or comparison with any 3.0 Three universal categories: firstness, secondness other phenomena (EP2, 150). As for its formal simplicity, the and thirdness category is derived from monadic predicates in the mathematical logic of relations (as distinguished from dyadic and Peirce’s most basic philosophical science, phenomenology, triadic predicates or relations).28 Both from its formal and ex- analyses universal elements in human experience, and he in- periential sources, its content is fixed ahead of any specialised troduces the categories “firstness,” “secondness” and “third- research into psychological correlates of the category, and it is ness” to describe and distinguish these elements. While prior to any distinctions introduced by other philosophical Peirce’s categories have an historical point of departure in the disciplines. Yet, as derived from ordinary experience, the phe- table of categories in Kant’s first Critique,26 he further evokes nomenological analysis provides further specifications of an Aristotelian-scholastic tradition by qualifying his catego- firstness in terms of experiential content. The category is thus 272 Knowl. Org. 47(2020)No.3 T. Midtgarden. Peirce’s Classification of the Sciences exemplified by “the quality of redness” as such, regardless of gories.38 From a formal point of view, Peirce’s derivation of its “vidness and dimness,” and taken as “not relative to any- the category from a mathematical logic of relations is intrin- thing” (EP2, 268). Firstness is further qualified through a sically connected to his argument for the irreducibility of “myriad-fold varieties” of sensory modalities: “an odor, say a triadic relations.39 Since thirdness is “further away from smell of attar, or … one infinite dead ache, [or] the hearing of common sense” (EP2, 158) a formal approach, therefore, [a] piercing eternal whistle” (EP2, 150). By emphatically ap- becomes more salient for the analysis of this category. Nev- pealing to pre-specialised experience, Peirce distinguishes ertheless, for phenomenology it is crucial that claims con- such modes of presentation from what psychology would cerning the universality and irreducibility of the categories make an object of research.29 On the other hand, like his two can be supported by considerations from ordinary or pre- other categories, firstness informs the sign classification de- specialised experience. As for experiential specifications of veloped by speculative grammar.30 thirdness, he sometimes considers examples of what he The category secondness is formally derived from the no- qualifies as “a triadic fact,” such as that of a person giving tion of dyadic relations in the mathematical logic of rela- something to someone and thus transferring ownership to tions.31 Abstaining from the latter’s technical specifications, the other person (EP2, 171 emphasis original). Such a fact, however, Peirce’s second category captures a rudimentary Peirce stresses, is constituted by a relation between all three idea of a pair of reacting singulars.32 In Peirce’s phenomeno- elements involved and cannot be reduced to relations in- logical analysis, secondness presupposes firstness, but can- volving only two of them. However, Peirce’s science classi- not be reduced to firstness.33 Peirce sometimes contrasts fication leaves open further possibilities for experiential secondness to firstness by appealing to “the element of specifications. Since all philosophical disciplines are based Struggle” or “a sense of resistance” experienced, for example, on ordinary experience, thirdness could be more indirectly through “making a strong muscular effort, say that of press- analysed and specified through other philosophical disci- ing with all your might against a half-open door” (EP2, plines, such as speculative grammar or general metaphysics, 150). Yet his phenomenological analysis is careful not to or the more specific metaphysical inquiries into principles qualify secondness in terms of a physical or material domain for natural and social science, respectively. as such. One reason for this may be found in the hierarchical Firstly, the formal derivation of thirdness provides con- order of the science classification. Being hierarchically prior ceptual input for a definition of what a “sign” is that serves to a metaphysics that sustains and justifies a division be- as a starting point for speculative grammar. Peirce explicates tween natural and social science, the phenomenological ac- the very idea of a sign in terms of an irreducible triadic rela- count of secondness must abstain from using received onto- tion between three correlates: a representation (or “repre- logical dichotomies to qualify this category; say, those of sentamen”), the object represented, and the interpretation body/mind, physical/psychological, non-human/human, (or “interpretant”) of the representation as representing the or effective/final causation. Peirce rather takes resort to same object (CP 2.242). He sometimes defines thirdness more concrete characterisations, such as “a sense of re- simply as such a triadic relation.40 His formal analysis of var- sistance,” but he also introduces more abstract distinctions, ious types of triadic relations provides for a general classifi- such as that between agent and patient.34 cation of a variety of linguistic as well as non-linguistic signs, Like firstness, the second category, too, is used to define including the forms of indexicality briefly mentioned forms of signs analysed and classified by speculative grammar. above.41 Through this sign classification, the three catego- In particular, secondness defines indexical signs and the in- ries gain significance for Peirce’s classification of the sci- dexicality of verbal discourse. The category enables separate ences as a whole. The categories thus inform and enable and step-wise considerations of, for example, dyadic and dy- considerations of signs essential to logic, such as different namic relations involved by attention raising indices (such as forms of proposition,42 and signs on which specialised scien- tones of voice),35 dyadic relations defining cognitive connec- tific observation and theorisation would depend, such as di- tions established between indices and indicated objects agrams,43 as well as types of scientific reasoning distin- (pointing gestures)36 and dyadic relations involved in furnish- guished semiotically through different kinds of sign inter- ing verbally mediated information about an indicated object pretants as abduction, induction and deduction.44 (propositional symbols).37 Furthermore, secondness, as well Peirce takes the categories, in particular thirdness, to sus- as firstness, provides resources for considering non-cognitive tain and enable conceptualisation and theoretisation in and non-linguistic ways in which human agents are embod- both natural and social science. In his hierarchical science ied in the world of experience, and for considering how such classification, it is the task of metaphysics to apply the cate- embodiment forms a basis for cognitive activity. gories in exploring general features of reality to provide an The analysis of thirdness concludes the categorical anal- ontological basis for theoretisation. As for the natural sci- ysis, and the third category presupposes firstness and ences, Peirce stresses that metaphysics would be concerned secondness but is reducible to neither of the two other cate- with the “reality of Thirdness” by qualifying the latter in Knowl. Org. 47(2020)No.3 273 T. Midtgarden. Peirce’s Classification of the Sciences terms of the reality of laws or regularities in nature (EP2, are expressively given through the contract’s intent “that 181). Assuming that “general principles are really operative certain conditional rules shall govern the conduct of [the in nature” (EP2, 183), his metaphysics thus advocates a ver- contractors]” (CP 1.475). To use a speech act theoretical sion of scholastic realism. By using rather simple examples distinction, in either case, language use should be analysed within the reach of ordinary experience, however, he asks in terms of “illocutionary effects,” rather than merely as what the reality of dispositional properties of inorganic “perlocutionary effects.”49 Like Peirce’s interpretant illocu- things would consist in, such as the hardness of a diamond. tionary effects can only be accounted for by triadic rela- This dispositional property, he argues, cannot be reduced to tions,50 while perlocutionary effects may be analysed as dy- actual observable states of reacting singular phenomena,45 adic relations between linguistic acts and certain physical, which would be instances of secondness. Rather, he points physiological or psychological reactions to those acts. Given out, the reality of dispositional properties proves itself these specifications, Peirce’s point about the irreducibility through what would happen under certain conditions. A of triadic relations and thirdness gains relevance for recent diamant’s hardness thus consists in that the diamond would developments in the social sciences, such as Jürgen Haber- sustain the pressure of a knife-edge without being scratched, mas’ (1984-7) theory of communicative action. if the knife-edge were drawn over it.46 By considering how thirdness defines ontological princi- Peirce’s explorations of thirdness further suggests the rel- ples in social and natural science and how Peirce’s three cat- evance of this category for social science research. More spe- egories inform his sign classification, we have noted how the cifically, in considering the triadic sign-relation, he stresses categories gain significance for his science classification as a how the triadic relation of the sign (or “representamen”) to whole. Yet, Peirce further suggests that his categories can be both an object and an interpretation (“interpretant”) is irre- applied directly to the divisions between the sciences in ducible. In particular, he qualifies this irreducible triadic re- qualifying both their distinctness and their hierarchical de- lation as an “intellectual fact” (EP2, 171), and he states more pendence relations. In distinguishing the three main philo- elaborately that a “three-subject fact is comprehensible and sophical sciences, he qualifies them briefly in terms of first- is analogous to an utterance, a speech” (CP 6.323). In more ness, secondness and thirdness, respectively. As for phenom- specific terms, however, an interpretation of a verbal utter- enology, he uses the category of firstness in a certain meta- ance is further defined and conditioned by shared norms scientific sense in pointing out that this science treats of and expectations. This is shown by Peirce’s example of how phenomena “in their immediate phenomenal character, in acts of asserting and assenting to a proposition are made un- themselves as phenomena … [and] … thus … in their First- der a mutual understanding that telling a lie would be met ness” (EP2, 197). Moreover, in attempting to qualify both with moral, and sometimes legal, sanctions.47 Such instances the distinctness of the normative sciences and their depend- may suggest how language use involves irreducible triadic re- ence on phenomenlogy, he uses the category of secondness lations to socially or institutionally conditioned interpreta- in a rather abstract and highly generalised sense. The nor- tions (interpretants), as well as to objects. On the other mative sciences, he claims, concern “the conformity of phe- hand, by neglecting such social and normative conditions, nomena to ends which are not immanent within those phe- the analysis might tend to account for sign use in terms of nomena” (EP2, 199 emphasis original), and in so far, they dyadic relations defined, for example, by an utterance and treat “of Phenomena in their Secondness” (EP2, 197). some psychological intention behind the former, or, alter- While admitting that this qualification might be “too nar- natively, by a semantic relation between utterance’s verbal row” (EP2, 197), his further qualification of metaphysics is meaning and its reference. An even more convincing case in line with our consideration above of its relevance for the for the irreducibility of thirdness is provided by Peirce’s ex- specialised sciences (EP2, 197): “Metaphysics … treats of ample of two persons making a contract.48 Such institution- Phenomena in their Thirdness.” Nevertheless, his use of ally bounded language use cannot be meaningfully analysed firstness and secondness in accounting for phenomenology, only in terms of sets of dyadic relations between reacting normative science and metaphysics, as well as for metaphys- singular phenomena specified as, say, a signer and a docu- ics’ dependence on the two former sciences, may seem little ment, between the two parties to a contract, or between the convincing compared to his more elaborate attempt to signed document and successive acts of the signers. Rather, ground both metaphysics and the specialised sciences in a a triadic relation would define the institutionally condi- sign classification and logical and methodological principles tioned intent of the contract (interpretant), the contract provided by speculative grammar, logical critic and (sign) and the state of affairs in the world the contract is methodeutic, respectively. about and is to regulate (object). As with an informal act of However, the categories may also be seen to bear on the assertion, the signing of a contract is binding on language subdivisions within each branch of specialised science. As users and has practical consequences if the norms at stake mentioned above, Peirce classifies both natural and the so- are broken. In the case of contracts, however, such norms cial sciences hierarchically into nomological, classificatory 274 Knowl. Org. 47(2020)No.3 T. Midtgarden. Peirce’s Classification of the Sciences and descriptive sciences.51 Thirdness informs and sustains and specialised sciences generally. As Peirce stresses time and the nomological sciences through laws or ontological prin- again, observational activities not only provide an empirical ciples for the various fields of natural and social science re- basis for confirmation or refutation of explanatory theories search, while these laws and principles are in turn to provide but is a main driver for scientific and theoretical develop- the basis for classification in the classificatory sciences and ment, and hence for scientific specialisation. What distin- for possible explanation of phenomena described by de- guishes specialised sciences as such, he points out, is “a pe- scriptive sciences. Yet, through this hierarchical order of culiar form of observation” (CP 1.99). Historically, new classification, Peirce deliberatively idealises actual scientific technologies of observation have enabled researchers to dis- developments and tendencies toward theorising laws and cover and take account of ever more fine-grained or com- regularities. In fact, the hierarchical order “reverses” the his- plex aspects of micro or macro level phenomena. “The great torical order of development in so far as there is “a well- landmarks in the history of science,” he notes, “are to be marked tendency for a science to be first descriptive, later placed at the points of where new instruments, or other classificatory, and lastly to embrace all classes in one law” means of observation, are introduced” (CP 1.102). One (CP 1.226). Moreover, Peirce’s historical exemplifications may thus distinguish “[a]stronomy before the telescope and provide opportunity to reflect on the distinctness, as well as astronomy after the telescope. Prephotographic astronomy the historical priority of “classificatory” and “descriptive sci- and photographic astronomy. Chemistry before the exact ences” in categorical terms and in terms of secondness and analytic balance, and after” (CP 1.102). While the phenom- firstness in particular. enal qualities mediated by sophisticated optical technolo- Distinguishing classificatory sciences more carefully, gies are different from those studied by phenomenology Peirce uses chemistry as an example, a science he knows through everyday life experience, they would still be phe- from training. Peirce admits that, in its actual state of devel- nomenal qualities, or, when qualified by Peirce’s semiotic opment, chemistry is little concerned with general physical distinctions, “qualisigns” or “icons” (CP 2.254). In so far, or physio-chemical laws but is “limited to the study of reac- they may be defined in terms of firstness. In fact, optical in- tions, to the structure of compounds, and to the behavior struments may be specifically designed to capture certain of elements in combination” (CP 1.260). Rather than in- phenomenal qualities while ignoring others, as seen, for ex- quiring deeper into the properties of the different sub- ample, in that “the qualities striven for in a telescopic objec- stances, a chemist seeks “to identify them and to make out tive are of no consequence in a microscopical objective” their constitutional relations” (CP 1.260). Hence, in em- (EP2, 131). Moreover, due to particular technologies of ob- phasizing concrete reactions and behaviors in identifying el- servation and particular training, a member of a certain sci- ements in combinations, Peirce’s account suggests that entific community, say, “a bacteriologist,” “will live in quite secondness is an underlying category in the chemist’s con- a different world - quite a different aggregate of experience,” ceptual and classificatory scheme. Secondness is further the than a layperson, but also than a member of a different sci- defining category in Peirce’s semiotic account of a chemist’s entific communty, and “neither has seen the world in which procedures for identifying a particular element, using lith- the other lives” (EP2, 131). Hence, in qualifying specialised ium as an example. A chemist identifies this element by us- sciences in terms of what the Husserlean phenomenologist ing an operational definition or a “precept” that “tells you would term “life worlds,”52 Peirce suggests that firstness, not what to do in order to gain a perceptual acquaintance with only secondness and thirdness, may distinguish specialised the object of the word” (CP 2.330). Since this definition in- sciences as such. structs the scientist how (s)he is to interact with and manip- ulate substances in order to identify the element, it works 4.0 The relevance of Peirce’s classification basically like an indexical, which Peirce defines in terms of secondness. Hence, using chemistry as a case, Peirce’s distin- Like Comte, Peirce stresses the unity and interdependence guishes what classificatory sciences are by giving priority to of the sciences, as well as their diversifications and their dis- secondness rather than thirdness. tinctness. While we have seen how Peirce’s categories of In Peirce’s hierarchical and Comte-inspired classifica- firstness and secondness may qualify the distinctness of the tion, descriptive sciences, such as geology and astronomy various specialised sciences, his category of thirdness, (CP 1.198), are subordinated to nomological and classifica- through his “nomological sciences,” would stress tendencies tory sciences. While “going into the utmost detail, describ- across the specialised sciences to search for general laws and ing individual phenomena” (EP2, 258), these sciences regularities. In particular, we noted that his division be- would endeavor “to explain their phenomena by the princi- tween the natural sciences and the social or human sciences ples of nomological and classificatory physics” (EP2, 259). in effect resists Dilthey’s dualistic scheme in taking account This hiearchical relationship, however, takes little account of the fact that parts of the humanities, too, are searching of how observational activities distinguish these sciences for laws or law-like regularities, as in the case of linguistics. Knowl. Org. 47(2020)No.3 275 T. Midtgarden. Peirce’s Classification of the Sciences

In this regard, Peirce’s classification gains relevance in the courses in logic, methodology and scientific writing.54 The light of Rens Bod’s (2013) recent comprehensive effort to stress on the interdependence of the sciences in Peirce’s clas- rewrite the history of the humanities.53 Contesting the sification gains further relevance when taken together with Diltheyan understanding-explanation dichotomy, as well as his pragmatist view of science as “a pursuit of living men” Wilhelm Windelband’s distinction between nomothetic (CP 1.232) and as a “living historic entity” (CP 1.44). As and idiographic approaches to knowledge, Bod’s study of Jaime Nubiola (2005) has pointed out, in several of his his- historical developments across the humanities shows (2013, torical reflections, Peirce considers cross- and interdiscipli- 7) how “the search for patterns and principles in the human- narity as a condition for scientific development. The hiear- ities is a continuous tradition.” While Peirce’s classification chical relationships between nomological, classificatory and provides sweeping evidence against Dilthey’s and Windel- descriptive sciences in his classification may thus translate band’s schemes from the late nineteenth century, Bod’s pro- more dynamically into influences, loans and interactions ject provides a comprehensive historical account from an- between sciences. For example, Peirce thinks that those who tiquity up to the present that challenges and undermines study “different kinds of plants and animals cannot attain these schemes. any true understanding of taxonomic biology until they can As noted above, Peirce’s hierarchical classification most be guided by the discoveries of the physiologists” (CP strikingly distinguishes itself from that of Comte by includ- 1.226). In turn, “the physiologist may be aided” by facts col- ing and assigning distinct roles to philosophical disciplines. lected by taxonomic biologists (CP 1.226). Moreover, sig- Peirce’s classification thus stresses the unity and interde- nificant scientific developments have been achieved pendencies of the sciences through principles articulated through interaction between the various disciplines and and analysed by phenomenology, the normative sciences fields of research: “Darwin adapted to biology the methods and metaphysics. Given Peirce’s elaboration of his norma- of Malthus and the economists; Maxwell adapted to … elec- tive sciences more specifically, we may highlight the rele- tricity the methods of hydrodynamics … Wundt adapts to vance of this branch of philosophy in particular. psychology the methods of physiology; … Cournot adapted Although Peirce fails to suggest ethical guidelines for the to political economy the calculus of variations” (EP1, 212). conduct of research in the contemporary sense, he does as- Extrapolating from these historical cases, Peirce ponders sign a superordinate role to ethics in relation to all special- that, in the future, major advances will come from cross-dis- ised empirical sciences. More specifically, in his science clas- ciplinary uptakes through which one succeeds “in adapting sification, ethics informs principles of “self-controlledˮ log- the methods of one science to the investigation of another” ical reasoning and also, more indirectly, the methodological (EP1, 212). In line with this hypothesised trajectory, we may application of logical reasoning in specialised empirical re- briefly note how comprehensive research fields have search. More importanly, however, in so far as ethics studies emerged through cross- or interdisciplinary uptake in the principles of right conduct in general and “what the fitness twentieth century, such as biochemistry, cellular biology of an ideal of conduct consists in” (CP 1.600), Peirce’s sci- and ecology in the natural sciences; neuropsychology, evo- ence classification would sustain and support a further ex- lutionary psychology and social psychology in psychology; ploration and specification of ethical norms for the conduct and socio-linguistics, gender studies, digital and environ- of research. Notably, his later proposal to extend the role of mental humanities in the social science and the humanities. the normative sciences to critically reflect on our inherited In particular, Peirce could be seen to anticipate the emer- moral beliefs and their applicability to “the new world cre- gence of the interdisciplinary field of science and technol- ated by science” (CP 5.513) seems even more relevant now ogy studies (STS) in stipulating “ethnology of technology” than in Peirce’s days. In particular, since the objective of this (EP2, 261) as part of ethnology. Yet some such develop- critical reflection would be to reconsolidate moral beliefs ments would challenge his dual branching of the specialised that “remain indubitable” (CP 5.513), Peirce’s proposal is sciences, in particular developments involving not only up- relevant for ongoing discussions of how our inherited moral take of methods but theories from other disciplines. Obvi- ideals of human dignity and worth bear on current medical ous examples would be evolutionary psychology and neuro- and genetic research, notably stem cell research. psychology, which would have to be taken as natural science Furthermore, Peirce’s semiotic account of logic may be disciplines in Peirce’s classification, rather than as psycho- seen to have not only theoretical but educational relevance. logical disciplines proper. Noting that his three semiotic disciplines speculative gram- At times Peirce stresses a kind of interdisciplinary inter- mar, logical critic and methodeutics is modelled on the me- action that in effect bridges an overarching division in his dieval trivium (grammar, logic and rhetoric) (EP2, 327), his classification: the division between science of discovery and semiotic account of logic may be seen to support university practical science. Interaction across this divide may evolve, curricula that aspire to establish a common frame of refer- he suggests, from issues or problems in the practical sciences ence for its various specialised studies through general and thus from some “dynamical relations” or “a compulsive 276 Knowl. Org. 47(2020)No.3 T. Midtgarden. Peirce’s Classification of the Sciences quality of action” (CP 7.52). Action coordination and co- 2. See Dipert 2004. operation arise in so far as “one group may stimulate an- 3. The historical point of departure of philosophical prag- other by demanding the solution of some problem” (CP matism in America is Peirce’s two texts “The fixation 7.52). Yet the interdisciplinary nature of such cooperation of beliefˮ (EP1, 109-23), originally published in 1877, would not consist simply in applications or practical adap- and “How to make our Ideas clearˮ (EP1, 124-41), tations of existing theories or methods in the specialised sci- published in 1878. ences. Peirce rather suggests that new theories and concepts 4. In this article, I draw both on Peirce’s classification of need to be developed in response to a major practical or so- the sciences from 1902 (which is found in “A Detailed cial problem, using as one of his examples Pasteur’s discov- Classification of the Sciencesˮ from “Minute Logicˮ eries of the nature of bacteria “with their far-reaching bless- [CP 1.203-83]) and his classification from 1903 (“An ings to the human species” (7.52). “In this way,” he main- outline classification of the sciencesˮ [EP2, 258-66]). tains, “the practical sciences incessantly egg on researches In so far as there are discrepancies between the two clas- into theory” (CP 7.52). The kind of theory and knowledge sifications, I follow the latter. For a developmental ac- development considered can thus be seen to bear similarity count of the two classifications, see Atkins (2006). to what Gibbons, Limoges, Nowotny et al (1994) and 5. See EP2, 35, EP2, 258. Nowotny, Scott and Gibbons (2001) more recently have 6. See CP 1.184, EP2, 144, 146. qualified as the production of “Mode 2” knowledge: 7. See Murphey (1961, 1-3). knowledge that is transdisciplinary, rather than disciplinary, 8. See Midtgarden (2001, 2007). socially distributed across various sectors of society and de- 9. See his “On a new list of categoriesˮ (EP1, 1-10). veloped in contexts of application and use. However, re- 10. See Atkins (2006, 487-8). calling his general problematisation of the extensive social 11. See CP 1.246, 1.249, 1.278. implementation of science and technology and the ensuing 12. Peirce’s classification of practical science is elaborated doubts concerning our inherited moral beliefs (CP 5.513), in MS 1343. Peirce could be seen to call for interdisciplinary involvement 13. See CP 1.234. also on part of the more general sciences of discovery, the 14. See also de Waal (2005). normative disciplines ethics, esthetics and logic in particu- 15. See “The Basis of Pragmaticism in Phaneroscopyˮ lar. In so far as scientific discoveries lead to technological in- (EP2, 360-70). terventions in industry, transportation or medicine that 16. See CP 1.241, EP2, 147-8. challenge basic moral beliefs, say, concerning human dignity 17. See Schmaus, Pickering, Bourdeau (2018). and welfare or environmental protection and animal wel- 18. See CP 5.513 and Midtgarden (2014). fare, Peirce’s proposal would invite ethical, esthetic and se- 19. See EP2, 326-30. miotic inquiries into the meaning and applicability of our 20. See also Midtgarden (2001). shared moral beliefs. Following Peirce’s proposal, such in- 21. See EP2, 39, EP2: 260. terdisciplinary engagement may further enhance moral re- 22. See EP2, 250, 264. flection in practitioners of relevant practical sciences and 23. See Peirce’s and Joseph Jastrow’s “On small differences forge a sense of moral and social responsibility in specialised of sensationˮ (1885) (reprinted in CP 7.21-35). For a empirical researchers involved. Hence, again the divisons thorough discussion of Peirce’s contribution to psy- and hiearchical dependencies in Peirce’s classification may chology, see Cadwallader (1975). be conceived as ports of entry for interaction and coopera- 24. See in particular EP2, 470-1. Peirce’s position may thus tion across the sciences. be seen as reflecting that of Wundt himself and in effect as being opposed to the efforts of the younger German Notes psychologists Oswald Külpbe, Hermann Ebbinghaus and E.B. Titchener of construing psychology as a natu- 1. References to Peirce’s works are to: 1) Collected Papers ral science. For a useful discussion of these latter efforts, of Charles Sanders Peirce (Peirce 1931–58) with the ab- see Danziger (1979). breviation “CP” and where the number to the left of 25. See EP2, 36-7. the decimal point indicates the volume and the number 26. See EP2, 148, CP 1.563. to the right the paragraph; 2) The Essential Peirce 27. See his “Nomenclature and Divisions of Triadic Rela- (Peirce 1992-8) with the abbreviation “EP,” and where tions, as far as they are determinedˮ (EP2, 289-99). references are to the volume and page number; 3) Un- 28. See EP2, 427-8. published manuscripts in microfilm version that are in- 29. Sometimes he seeks to avoid the term “feeling” and ra- dicated with “MS.,” and the numbers refer to those in ther defines firstness through “sense of quality,” which Robin 1967. would, for example, be “the sort of element that makes Knowl. Org. 47(2020)No.3 277 T. Midtgarden. Peirce’s Classification of the Sciences

red to be such as it is, whatever anything else may be” furthermore, he knows himself to be included with all (CP 8.267). his activities and all his theoretical ideas” (1970, 121). 30. In one of Peirce’s most systematic classification of 53. I thank Rasmus Slaattelid for drawing my attention to signs, he uses firstness (although not the word “first- Bod’s research. ness” in this context but rather “first” and “possibil- 54. A case in point is the introductory course Examen Phil- ity”) in a technical and step-wise manner to define the osophicum at Norwegian Universities. This course is syntax of a propositional symbol, a so-called “dicent mandatory for all curricula and exams at, for example, symbol” (EP2, 295-6). Such a symbol requires through the University of Oslo (see Lovdata 2006). its interpretation (or “interpretant”) an “iconic legisign,” where the latter has already been exemplified by References “a diagram, a part from its factual individuality” (EP2, 294), and “icon” has been defined by firstness (EP2, Ambrosio, Chiara. 2016. “The Historicity of Peirce’s Clas- 291) and “legisign” by thirdness (EP2, 291). Yet, even sification of the Sciences.ˮ European Journal of Pragma- firstness has here initially been defined as an element of tism and American Philosophy 8, no. 2. doi:10.4000/ triadic relations (EP2, 289-90). ejpap.625 31. For Peirce’s algebraic treatments of dyadic relatives, see Atkins, Richard Kenneth. 2006. “Restructuring the Sci- for instance CP 3.492-8. ences: Peirce's Categories and His Classifications of the 32. See EP2, 160-1. Sciences.ˮ Transactions of the Charles S. Peirce Society 42, 33. CP 1.530, EP2, 270. no. 4: 483-500. 34. EP2, 150. Austin, John L. 1962. How to Do Things with Words. Ox- 35. In Peirce’s terminology, this would be a “rhematic in- ford: Oxford University Press. dexical sinsign” (EP2, 294). Bentham, Jeremy (1817) 1983. Chrestomathia, ed. M. J. 36. This would be what Peirce calls a “dicent sinsign” (EP2, Smith and W. H. Burston. Oxford, Clarendon Press. 294). Bod, Rens. 2013. A New History of the Humanities: The 37. This would be a “dicent symbol” (EP2, 295). Search for Principles and Patterns from Antiquity to the 38. CP 1.530, EP2, 240. Present. Oxford: Oxford University Press. 39. For a discussion of Peirce’s argument, see Hookway Cadwallader, Thomas C. 1975. “Peirce as an Experimental 1985: 97-101. Psychologist.ˮ Transactions of the Charles S. Peirce Society 40. See CP 1.541. 11, no. 3: 167-86. 41. See Peirce’s ten-fold classifications of signs, EP2, 289- Comte, Auguste. 1855. The Positive Philosophy of Auguste 99. Comte, trans. Harriet Martineau. New York, NY: C. 42. See EP2, 299. Blanchard. 43. EP2, 294, 298. Danziger, Kurt. 1979. “The Positivist Repudiation of 44. EP2, 297-9. Wundt.ˮ Journal of the History of the Behavioral Sciences 45. See EP2, 396, 401-2. 15: 205-30. 46. See EP2, 354, 356-7. de Waal, Cornelis. 2005. “Why Metaphysics Needs Logic and 47. See EP2, 278, 311-12, CP 5.546. Mathematics Doesn't: Mathematics, Logic, and Meta- 48. CP 1.475. physics in Peirce's Classification of the Sciences.ˮ Transac- 49. See Austin (1962, 101ff). tions of the Charles S. Peirce Society 41, no. 2: 283-97. 50. In agreement with Peirce’s triadic sign-relation, an illo- Dilthey, Wilhelm. 1989. Introduction to the Human Sci- cutionary act would involve a socially (or institution- ences. In Wilhelm Dilthey: Selected Works Vol. I, ed. Ru- ally) conditioned uptake (interpretant), an uttered dolf A. Makkreel and Frijhthof Rodi. Princeton, NJ: propositional symbol or “locutionary act” (sign), the Princeton University Press. propositional content of which would be about some- Dipert, Randall. 2004. “Peirce’s Deductive Logic: Its Devel- thing in the world (object). As for the latter, however, opment, Influence, and Philosophical Significance.ˮ In an illocutionary act would involve either a “world-to- The Cambridge Companion to Peirce, ed. Cheryl Misak. word,” or a “word-to-world direction of fit” (see Searle Cambridge: Cambridge University Press: 287-324. 1979, 11-20). Gadamer, Hans-Georg. 2004. Truth and Method, trans. 51. EP2, 259-62. Joel Weinsheimer and Donald G. Marshall. New York, 52. Note how Husserl sometimes qualifies his concept of NY: Continuum. Translation of Wahrheit und Metho- the life world in terms of “the world in which [the phys- de. icist] sees his measuring instuments, hears time-beats, Gibbons, Michael, Camille Limoges, Helga Nowotny et al. estimates visible magnitudes, etc. - the world in which, 1994. The New Production of Knowledge: The Dynamics 278 Knowl. Org. 47(2020)No.3 T. Midtgarden. Peirce’s Classification of the Sciences

of Science and Research in Contemporary Societies. Lon- with a preface by Cornelis de Waal. Semiotics, Commu- don: Sage Publications. nication and Cognition 14. Boston: De Gruyter Mou- Habermas, Jürgen. 1984-7. Theory of Communicative Ac- ton, 421-7. tion, trans. Thomas A. McCarthy. Boston, MA: Beacon Murphey, Murray G. 1961. The Development of Peirce’s Phi- Press. Translation of Theorie des kommunikativen Han- losophy. Cambridge, MA: Harvard University Press. delns. Nowotny, Helga, Peter Scott and Michael Gibbons. 2001. Heidegger, Martin. 2008. Being and Time, trans. John Ma- Re-Thinking Science: Knowledge and the Public in an Age quarrie and Edward Robinson. New York, NY: Harper of Uncertainty. Cambridge: Polity Press. Perennial. Translation of Sein und Zeit. Nubiola, Jaime. 2005. “The Classification of the Sciences Hookway, Christopher. 1985. Peirce. London: Routledge and Cross-Disciplinarity.ˮ Transactions of the Charles S. and Kegan Paul. Peirce Society 41, no. 2: 271-82. Husserl, Edmund. 1970. The Crisis of European Sciences Peirce, Charles Sanders. 1931-58. Collected Papers of Charles and Transcendental Phenomenology: An Introduction to Sanders Peirce, ed. Charles Hartshorne and Paul Weiss Phenomenological Philosophy, trans. David Carr. Ev- (vols. I-VI, 1931-5), Arthur Burks (vols. VII-VIII, 1958). anstone: Northwestern University Press. Translation of Cambridge, MA: Belknap Press. Krisis der europäischen Wissenschaften und die transzen- Peirce, Charles Sanders. 1992-8. The Essential Peirce: Selected dentale Phänomenologie. Philosophical Writings, ed. Nathan Houser and Christian Kant, Immanuel. 1998. Critique of Pure Reason, trans. and J. W. Kloesel (vol. 1), the Peirce Edition Project (vol. 2). 2 ed. Paul Guyer and Allen W. Wood. Cambridge: Cam- vols. Bloomington, IN: Indiana University Press. bridge University Press. Translation of Kritik der reinen Robin, Richard S. 1967. Annotated Catalogue of the Papers Vernunft. of Charles S. Peirce. Amherst, MA: University of Massa- Lovdata. 2006. “Forskrift om studier og eksamener ved Uni- chusetts Press. versitetet i Oslo” https://lovdata.no/dokument/SF/for- Schmaus, Warren, Mary Pickering, and Michel Bourdeau. skrift/2005-12-20-1798 2018. “Introduction.” In Love, Order, and Progress: The Midtgarden, Torjus. 2001. “Peirce's Speculative Grammar Science, Philosophy, and Politics of Auguste Comte, ed. War- from 1895-1896: Its Exegetical Background and Signifi- ren Schmaus, Mary Pickering, and Michel Bourdeau. cance.” Transactions of the Charles S. Peirce Society 37, Pittsburgh, PA: University of Pittsburgh Press: 3-24. no. 1: 81-96. Searle, John R. 1979. Expression and Meaning. Cambridge: Midtgarden, Torjus. 2007. “Peirce’s Epistemology and Its Cambridge University Press. Kantian Legacy: Exegetic and Systematic Considera- Whewell, William. 1847. “Classification of Sciences.” In The tions.” Journal of the History of Philosophy 45, no. 4: 577- Philosophy of the Inductive Sciences, Founded Upon Their 602. History. 2nd ed. London: John W. Parker, v. 2: 117-8. Midtgarden, Torjus. 2014. “Pragmatism, Cultural Lags and Whitehead, Alfred North, Russell, Bertrand. 1910-13. Moral Self-Reflection.” In Charles Sanders Peirce in His Principia Mathematica. Cambridge: Cambridge Uni- Own Words: 100 Years of Semiotics, Communication and versity Press. Cognition, ed. Torkild Thellefsen and Bent Sørensen

Knowl. Org. 47(2020)No.3 279 Books Recently Published

Books Recently Published Compiled by J. Bradford Young

DOI:10.5771/0943-7444-2020-3-279

Báu, TS. Nguyễn Văn, ThS. Nguyễn Thị Ly and ThS. Holzberlein, Deanne. 2020. Computer Software Cataloging: Nguyễn Phạm Ngọc Hân, eds. 2020. Những vấn đề lưu Techniques and Examples. London: Routledge. trữ học - quản trị văn phòng trong bối cảnh đất nước đổi Jackson, Sarah J., Moya Bailey and Brooke Foucault Welles. mới và hội nhập. Thành phố Hồ Chí Minh: Nhà xuất 2020. #Hashtagactivism: Networks of Race and Gender bản Đại học quốc gia Thành phố Hồ Chí Minh. Justice. Cambridge, MA: MIT Press. Bell, Hazel K. 2020. Indexing Biographies and Other Stories Liarokapis, Fotis, Athanasios Voulodimos, Nikolaos Dou- of Human Lives. 4th ed. Liverpool: Liverpool University lamis and Anastasios Doulamis, eds. 2020. Visual Com- Press. puting for Cultural Heritage. Cham: Springer. Biagioli, Mario and Alexandra Lippman, eds. 2020. Gaming Mansī, Isrāʼ ʻAbd Rabbih. 2020. Fahrasat al-mujassamāt the Metrics: Misconduct and Manipulation in Academic al-turāthīyah fī muʼassasāt al-maʻlūmāt: bayna qawāʻid Research. Cambridge, MA: MIT Press. wa-miʻyār. Dimashq: Nūr Ḥūrān lil-Dirāsāt wa-al-Nashr Byghan, Yowann. 2020. Sacred and Mythological Animals: wa-al-Turāth. A Worldwide Taxonomy. Jefferson, NC: McFarland. Paul, Dimple Valayil. 2020. Developing A Keyword Extrac- Carlson, Scott, Cory Lampert, Darnelle Melvin and Anne tor and Document Classifier: Emerging Research and Op- Washington. 2020. Linked Data for the Perplexed Li- portunities. Hershey, PA: Information Science Refer- brarian. Chicago: ALA Editions. ence. Cimiano, Philipp, Christian Chiarcos, John P. McCrae and Rocha, Zarine L. and Peter J. Aspinall, eds. 2020. The Pal- Jorge Gracia. 2020. Linguistic Linked Data: Representa- grave International Handbook of Mixed Racial and Eth- tion, Generation and Applications. Cham: Springer. nic Classification. Cham: Palgrave Macmillan. Flanders, Judith. 2020. A Place for Everything: The Curious Testa, Dean, Johel Brown-Grant and Denise Bedford. 2020. History of Alphabetical Order. London: Picador. Assessment Strategies for Knowledge Organizations. Hima Bindu, K., Raghava Morusupalli, Nilanjan Dey and Bingley, UK: Emerald. C. Raghavendra Rao. 2020. Coefficient of Variation and Yaneva, Albena. 2020. Crafting History: Archiving and the Machine Learning Applications. New York: CRC Press. Quest for Architectural Legacy. Ithaca, NY: Cornell Uni- Holeňa, Martin, Petr Pulc and Martin Kopp. 2020. Classi- versity Press. fication Methods for Internet Applications. Cham: Springer.

Knowl. Org. 47(2020)No.3

KNOWLEDGE ORGANIZATION KO

Official Journal of the International Society for Knowledge Organization ISSN 0943 – 7444 International Journal devoted to Concept Theory, Classification, Indexing and Knowledge Representation

Publisher Remove all active hyperlinks, including those from reference formatting software (if hovering over the text with a mouse produces a gray highlight, the text is hyperlinked; remove the link “Insert,” Ergon – ein Verlag in der Nomos Verlagsgesellschaft mbH “Hyperlink,” “Remove link”). Waldseestraße 3-5 Upon acceptance of a manuscript for publication, authors must provide a digital photo and a D-76530 Baden-Baden one-paragraph biographical sketch (fewer than 100 words). The photograph should be scanned with Tel. +49 (0)7221-21 04-667 a minimum resolution of 600 dpi and saved as a .jpg file. Fax +49 (0)7221-21 04-27 Sparkasse Baden-Baden Gaggenau IBAN: DE05 6625 0030 0005 0022 66 References BIC: SOLADES1BAD Reference citations within the text should have the form: (Author year). For example, (Jones 1990). Specific page numbers are required for quoted material, e.g. (Jones 1990, 100). A citation with two KNOWLEDGE ORGANIZATION authors would read (Jones and Smith 1990); three or more authors would be: (Jones et al. 1990). Journal of the International Society for Knowledge Organization When the author is mentioned in the text, only the date and optional page number should appear in Richard P. Smiraglia, Editor-in-Chief parentheses: “According to Jones (1990), …” or “Smith wrote (2010, 146): ….” A subsequent page [email protected] reference to the same cited work (e.g., to Smith 2010) should have the form “(229).” There is never a comma before the date. Instructions for Authors In-text citations should not be routinely placed at the end of a sentence or after a quotation, but an attempt should be made to work them into the narrative. For example: Manuscripts should be submitted electronically (in Microsoft® Word format) in English only via Jones (2010, 114) reported statistically significant results. ScholarOne at https://mc04.manuscriptcentral.com/jisko. Manuscripts that do not adhere to these Many authors report similar data; according to Matthews (2014, 94): “all seven studies report guidelines will be returned to the authors for resubmission in proper form. means within ±5%.” Manuscripts should be accompanied by an indicative abstract of approximately 250 words. In-text citations should precede block quotations, and never are placed at the end of a block- Manuscripts of articles should fall within the range 6,000-10,000 words. Longer manuscripts will be quotation. considered on consultation with the Editor-in-chief. References should be listed alphabetically by author at the end of the article. Reference lists A separate title page should include the article title and the author’s name, postal address, and E- should not contain references to works not cited in the text. Websites mentioned in passing in the text mail address. Only the title of the article should appear on the first page of the text. Contact infor- should be identified parenthetically with their URLs but not with references unless a specific page of mation must be present for all authors of a manuscript. a specific website is being quoted. To protect anonymity, the author’s name should not appear on the manuscript. Author names should be given as found in the sources (not abbreviated, but also not fuller than Criteria for acceptance will be appropriateness to the field of knowledge organization (see what is given in the source). Journal titles should not be abbreviated. Multiple citations to works by “Scope” at https://www.isko.org/ko.html), taking into account the merit of the contents and the same author should be listed chronologically and should each include the author’s name. Articles presentation. It is expected that all successful manuscripts will be well-situated in the domain of appearing in the same year should have the following format: “Jones 2005a, Jones 2005b, etc.” knowledge organization, and will cite all relevant literature from within the domain. Authors are Proceedings must be identified fully by volume title (which often differs from the name of the encouraged to use the KO literature database at http://www.isko.org/lit.html. conference), editor(s), series (if applicable), and details of publication (place and publisher name). Pa- The manuscript should be concise and should conform to professional standards of English us- pers in proceedings must be identified by their page numbers. Papers from online proceedings that age and grammar. Authors whose native language is not English are encouraged to make use of pro- lack publication details should include a DOI or URL. fessional academic English-language proofreading services. We recommend Vulpine Academic Ser- Journal issue numbers are given only when a journal volume is not through-paginated. Refer- vices ([email protected]). ences for published electronic resources should be accompanied by either a URL or DOI but not in Manuscripts are received with the understanding that they have not been previously published, lieu of actual publication data; access dates are not allowed. are not being submitted for publication elsewhere, and that if the work received official sponsorship, Unpublished electronic resources may use an access date in lieu of a data of publication. In cases it has been duly released for publication. Submissions are refereed, and authors will usually be notified of doubt, authors are encouraged to consult The Chicago Manual of Style 17th ed. (or online), author- within 6 to 8 weeks. date reference system (chapter 15).

Examples: Manuscript Formatting Dahlberg, Ingetraut. 1978. “A Referent-Oriented, Analytical Concept Theory for INTERCON-

Under no circumstances should the author attempt to mimic the presentation of text as it appears in CEPT.” International Classification 5: 142-51. our published journal. Instead, please follow these instructions: Howarth, Lynne C. 2003. “Designing a Common Namespace for Searching Metadata-Enabled In Microsoft® Word please set the language preference (“Tools,” “Language”) to “English (US)” Knowledge Repositories: An International Perspective.” Cataloging & Classification Quarterly or “English (UK).” 37, nos. 1/2: 173-85. The entire manuscript should be double-spaced, including notes and references. Pogorelec, Andrej and Alenka Šauperl. 2006. “The Alternative Model of Classification of Belles-Let- The text should be structured with decimally-numbered subheadings (1.0, 1.1, 2.0, 2.1, 2.1.1, tres in Libraries.” Knowledge Organization 33: 204-14. etc.). It should contain an introduction, giving an overview and stating the purpose, a main body, Schallier, Wouter. 2004. “On the Razor’s Edge: Between Local and Overall Needs in Knowledge Or- describing in sufficient detail the materials or methods used and the results or systems developed, and ganization.” In Knowledge Organization and the Global Information Society: Proceedings of the a conclusion or summary. Eighth International ISKO Conference 13-16 July 2004 London, UK, edited by Ia C. McIlwaine. Author-generated keywords are not permitted. Advances in knowledge organization 9. Würzburg: Ergon Verlag, 269-74. Footnotes are not allowed but endnotes may be used. Do not use automatic footnote formatting. Smiraglia, Richard P. 2001. The Nature of ‘a Work’: Implications for the Organization of Knowledge. Instead, insert a superscript numeral (Format, Font, Superscript) and create the text of the note man- Lanham, Md.: Scarecrow. ually in a separate list at the end of the manuscript, before the reference list. Smiraglia, Richard P. 2005. “Instantiation: Toward a Theory.” In Data, Information, and Knowledge Paragraphs should include a topic sentence, a developed narrative and a conclusion; a typical in a Networked World; Annual Conference of the Canadian Association for Information Science paragraph has several sentences. Paragraphs with tweet-like characteristics (one or two sentences) are … London, Ontario, June 2-4 2005, ed. Liwen Vaughan. http://www.cais-acsi.ca/2005proceed- inappropriate. ings.htm.

Italics are permitted only for phrases from languages other than English, and for the titles of pub- © Ergon – ein Verlag in der Nomos Verlagsgesellschaft, lished works. Baden-Baden 2020. All Rights reserved. Bold type is not permitted.

Em-dashes should not be used as substitutes for commas. Dashes must be inserted manually (In- KO is published by Ergon. sert, Advanced Symbol, Em-dash) with no spaces on either side. Illustrations should be embedded within the document. Photographs (including color and half- Annual subscription 2020: tone) should be scanned with a minimum resolution of 600 dpi and saved as .jpg files. Tables should – Print + online (8 issues/ann.; unlimited access for your Campus via Nomos eLibrary) contain a number and caption at the bottom, and all columns and rows should have headings. All € 359,00/ann. illustrations should be cited in the text as Figure 1, Figure 2, etc. or Table 1, Table 2, etc. – Prices do not include postage and packing Examples of KOS (classification, taxonomy, ontology, thesaurus, etc.) arrays should be config- – Cancellation policy: Termination within 3 months‘ notice to the end of the calendar ured as figures and set into the document as jpgs; they should not be entered as editable text. year Knowl. Org. 47(2020)No.3

KO KNOWLEDGE ORGANIZATION

Official Journal of the International Society for Knowledge Organization ISSN 0943 – 7444 International Journal devoted to Concept Theory, Classification, Indexing and Knowledge Representation

Scope Aims

The more scientific data is generated in the impetuous present times, the Thus, KNOWLEDGE ORGANIZATION is a forum for all those interested in the organization of knowledge on a universal or a domain-specific more ordering energy needs to be expended to control these data in a re- scale, using concept-analytical or concept-synthetical approaches, as well as trievable fashion. With the abundance of knowledge now available the quantitative and qualitative methodologies. KNOWLEDGE ORGANI- questions of new solutions to the ordering problem and thus of improved ZATION also addresses the intellectual and automatic compilation and classification systems, methods and procedures have acquired unforeseen use of classification systems and thesauri in all fields of knowledge, with significance. For many years now they have been the focus of interest of in- special attention being given to the problems of terminology. formation scientists the world over. KNOWLEDGE ORGANIZATION publishes original articles, re- ports on conferences and similar communications, as well as book reviews, Until recently, the special literature relevant to classification was pub- letters to the editor, and an extensive annotated bibliography of recent clas- lished in piecemeal fashion, scattered over the numerous technical journals sification and indexing literature. serving the experts of the various fields such as: KNOWLEDGE ORGANIZATION should therefore be available at every university and research library of every country, at every information philosophy and science of science center, at colleges and schools of library and information science, in the science policy and science organization hands of everybody interested in the fields mentioned above and thus also at every office for updating information on any topic related to the prob- mathematics, statistics and computer science lems of order in our information-flooded times. library and information science KNOWLEDGE ORGANIZATION was founded in 1973 by an in- archivistics and museology ternational group of scholars with a consulting board of editors represent- journalism and communication science ing the world’s regions, the special classification fields, and the subject areas industrial products and commodity science involved. From 1974-1980 it was published by K.G. Saur Verlag, Mün- terminology, lexicography and linguistics chen. Back issues of 1978-1992 are available from ERGON-Verlag, too. As of 1989, KNOWLEDGE ORGANIZATION has become the of- Beginning in 1974, KNOWLEDGE ORGANIZATION (formerly IN- ficial organ of the INTERNATIONAL SOCIETY FOR KNOW- LEDGE ORGANIZATION (ISKO) and is included for every ISKO- TERNATIONAL CLASSIFICATION) has been serving as a common member, personal or institutional in the membership fee. platform for the discussion of both theoretical background questions and Annual subscription 2020: Print + online (8 issues/ann.; unlimited ac- practical application problems in many areas of concern. In each issue ex- cess for your Campus via Nomos eLibrary) € 359,00/ann. Prices do not perts from many countries comment on questions of an adequate structur- include postage and packing. Cancellation policy: Termination within 3 ing and construction of ordering systems and on the problems of their use months‘ notice to the end of the calendar year in opening the information contents of new literature, of data collections Ergon – ein Verlag in der Nomos Verlagsgesellschaft mbH, Wald- and survey, of tabular works and of other objects of scientific interest. Their seestraße 3-5, D-76530 Baden-Baden, Tel. +49 (0)7221-21 04-667, Fax +49 contributions have been concerned with (0)7221-21 04-27, Sparkasse Baden-Baden Gaggenau, IBAN: DE05 6625 0030 0005 0022 66, BIC: SOLADES1BAD

Founded under the title International Classification in 1974 by Dr. (1) clarifying the theoretical foundations (general ordering theory/ Ingetraut Dahlberg, the founding president of ISKO. Dr. Dahlberg served science, theoretical bases of classification, data analysis and reduc- as the journal’s editor from 1974 to 1997, and as its publisher (Indeks Ver- tion) lag of Frankfurt) from 1981 to 1997. (2) describing practical operations connected with indexing/classifi- The contents of the journal are indexed and abstracted in Social Sciences cation, as well as applications of classification systems and thesauri, Citation Index, Web of Science, Information Science Abstracts, INSPEC, Li- manual and machine indexing brary and Information Science Abstracts (LISA), Library, Information Sci- ence & Technology Abstracts (EBSCO), Library Literature and Information (3) tracing the history of classification knowledge and methodology Science (Wilson), PASCAL, Referativnyi Zhurnal Informatika, and Socio- (4) discussing questions of education and training in classification logical Abstracts. (5) concerning themselves with the problems of terminology in general and with respect to special fields.