Ie

Contents

The Design of Subject Access Elements in Online Public Access Catalogs

The Effectiveness of a Searching Thesaurus in Free-Text Searching in a Full-Text Database

A Proposed Methodology [or Examining the Provision of Subject Access ill the 0 PAC

Establishing a Compatible General Vocabulary in China: The Capability International Classification VoI.17(1990) No.2 UDC 025.4:168:001.4 (05)

INTERNATIONAL Contents CLASSIFICATION

Devoted to Concept Theory,Systematic Tennino­ logy and Organization of Knowledge Editorial The journal is the organ of the INTERNATIONAL Knowledge Organization and the OPAC ...... 65 SOCIETY FOR KNOWLEDGE ORGANIZA­ TION. (General Secretariat: Woogstr. 36 a, 0-6000 Frankfurt 50) Articles Editors Godert, W., Horny,S.: The design of subject access elements in Dr. Ingetraut DAHLBERG (Editor-in-Chief), Woogstr. 63a, 0-6000 Frankfurt 50 online publicaccess catalogs ...... 66 Dr. Robert FUGMANN, Alte Poststr. 13, 0-6270 Kristensen, J., Jiirvelin, K.: The effectiveness of a searching Idstein thesaurus in free-text searching in a full-text database .. , ...... 77 Prof. Jean M. PERREAULT, The Library, Univer­ Furniss, P.: A proposed methodology for examining the provision sity of AJabamaat Huntsville,P.O.Box 2600, Hunts­ villc,AL 35807,USA of subject access in the OPAC ...... 85 ConsultingEditors Zeng, L.: Establishing a compatible general vocabularyin China: Kenneth BAKEWELL, Liverpool Polytechnic, rm the capability ...... 91 School of Info . Sci. & Techno!., 79 Tithebarn Street,Liverpool, L2 2ER, U.K. Prof. Pauline A.COCHRANE, 6568 Clagett Ave., Reports and Commnnications Fairhaven, MD, 20754,USA Nathalie DUSOULIER,Inst. de l'Infonn. Science et Pertinent Research at OCLC ...... 99 Techn., Chateau de Montet, F-54500 Vandoeuvre, Meeting on Concept Relationships ...... 76 Nancy,France Information, Data, Knowledge, Classification and Structuring ...... 98 Douglas FOSKETT, 1,Dale Side, Gerrards Cross, Bucks. SL97JF, U.K. Preliminary Program TKE' 90 ...... 100 Jens B. FRITS-HANSEN, Hoesterkoebvej 7, Towards the Millenium. Library Catalogs for the Nineties ...... 109 DK-2970 Hoersholm,Denmark 2nd Summer School for Information Science, Konstanz ...... 109 Alan GILCHRIST, 38 Ship Street, Brighton BNl lAB, U.K. Prof. Dip!. Math. Winfried GODERT, FH Bib!. u. ISKONews3 Dok., Claudiusstr. l, D-5000 K6in 1 First International ISKO Conference, Darmstadt,Aug.14-17, 1990- Dr. M. A. GOPINATH, DRTC, Indian Statist. • Inst., 8th Mile, Mysore Road, R.Y. College P.O., First ISKO Working Group Established - First Meeting of the Bangalore-5600059, India lSKOIGC W GIndexing, Classifying, and Retrieval-SCCAC News ... 101 Prof. Eric de GROLIER, ISSC, 1, rue Miollis, F-75015 Paris, France Prof. Dr. JosefHOLZL, Inst. f. Techn. u. Warenwirt­ FIDICR News 27 schaft., Wirtschaftsuniv., Franz-Klcin-Gasse 1, A-ll90 Wien Task Force on UDC System Development (102)- Prof. Krishan KUMAR, Dept. of Libr. & Inform. 5th International Study Conference on Classification Research- Sci.,University orDehli,Deh li-lI0007, India 1990 Conferences of Interest to FID/CR Members- International Prof. Yukio NAKAMURA, 14-8, Nisikata-l, Bunkyo-ku, Tokyo 113,Japan Symposium on "Terminology in Advanced Microcomputer Peter PAULSON, Forest Press OCLC, 85 Water­ Applications", Vienna, Nov.22-23, 1989-U.K.: CRG275-USA: SIG/CR vlietAve.,Albany, NY, 12206, USA Program for ASlS Mid-Year-ASIS'90 SIG/CR Progam Preview- Drs. Gerhard J.A. RIESTHUIS, Univcrsitetsbibli­ Library Classilication News ...... 104 otheek, Singe! 425, NL-1012 WP Amsterdam, Netherlands Leon ROLLING, CEC, Dir. XTIl, lean-Monnet Centre,Kirchberg, LU-Luxembourg Book Reviews Dr. Winfried SCHMITZ-ESSER, Oderfeldstr. 13, Koskiala, S., Launo, R.(Eds.): Information·-Knowledge­ D-2000 Hamburg 13 Evolution. Proc.44th FID Congress, Helsinki 1988. Prof. Dr. Eugeniusz SCIBOR, llNTE, Broniews­ (P.A.Cochrane) ...... lOS kigo 83-157, PL-0l-876 Warszawa, Poland Buchanan, Dr. Otto SECHSER, In der Ey 37,CH-8047 Zurich, B.: Bibliothekarische Klassifikationstheorie Switzerland (Theoryof Library Classification) (H.Schnelling) ...... 106 Prof. Dr. Dagobert SOERGEL, College or Libr. & Comaromi, J.P.(Ed.) et al: Dewey Decimal Classification and Tnform. Serv., University of Maryland, College Park, MD, 20742, USA Relative Index. Ed.20. (M.P.Satija) ...... 107 Sharma, P.S.K.: Dr. Edouard SUKIASJAN, Lenin State Library of Universe of Knowledge and Research the USSR,Pr. Kalinina3,Moskva- IOIOO,USSR Methodology . (M.P.Satija) ...... 108 Prof. Nancy WILLIAMSON, Fac. of Libr. & In­ form. Sci., 140 St.George Str., Toronto M5S IAI, Literatnre 17(1990)No.2 ...... 110 Canada UDC 025.4 + 168 + 001.4 (05) INTERNATIONAL CLASSIFICATION

Devoted to Concept Theory. Systematic Terminology and Organization of Knowledge

Editors: Dr. I. Dahlberg, Frankfurt Dr.R.Fugmann, Idstein: Prof. J.M. Perreault, Huntsville, Ala.. USA; Editorial Office:c/o Dr. I. Dahlberg, Woogstr. 36a, 0-6000 Frankfurt 50. Tel. 0 69 / 52 36 90; FAX 0 69 / 52 05 66 Issue frequency: 4x/ann. Annual subscription rates: Institution per volume OM 94,-; Individuals per volume 20 % less if directly ordered from IN­ DEKS Verlag. Single issues: DM 28,-, Back issues available, ask for special offer. Subscription included in membership fee for members afthe Interna­ tiona! Society for Knowledge Organization (ISKO). MwSt (domestic only) and postage are not included; for Ail Mail delivery an additional OM 4.­ per issue required. Advertising rates: Advertising rate card No.2. Advertising fact sheet with details of mechanical requirements and dosing date available upon request. Publisher: INOEKS Verlag, Woogstr. 36a, 0-6000 Frankfurt 50, Tel. 0 69(52 36 90; FAX 0 69 / 52 05 66. Bank Account Nos.: INOEKS Verlag. Postgiro-Konto Frankfurt. No. 151 208-608 (BLZ 500 100 60); Frankfurter Sparkasse. No. 852 082 (BLZ 500 502 01 ) All rights reserved. Printed in the Federal Republic of Germany 1989 by Oruckerei H. Guntrum II KG, PF 180, 0-6407 Schlitz!Hessen.

Contents page I.C. Vol.17 (1990) No.2

G6dert, W., Horny,S.: The design of subject access elements in Kristensen, J., Jarvelin, K.: The effectiveness of a searching the­ Online Public Access Catalogs. saurus in free-text searching in a full-text database. InLClassif.17(1990)No.2, p.66-76, 73 refs. InLClassif.17(1990)No.2, p.77-84, 30 refs.

Experiences with retrieval in OPACs suggest that there arc Synonyms, antonyms, quasi-synonyms and homonyms are typi­ major problems for the users of such systems, such as choosing cal features of natural language which cause difficulties in fr ee­ the 'correct' subject access vocabulary, narrowing or broade­ text searching in full-text databases. Searching thesauri which ning the set of items retrieved regarding the search interest. Such supply synonyms, antonyms and quasiMsynonyms for given problems relate to two facts: (I) that in the early OPACs subject natural language terms have been suggested as a search aid to access has not been considered seriously enough and (2) that the overcome these problems. To test these suggestions, a small complexity of the topic has not been really recognized. Using a searching thesaurus was constructed for Finnish newspaper ar­ typology of users' questions it is demonstrated which require­ ticles on economic issues and a set of test searches were con­ ments musts be met by a succcssful online subject access. Im­ ducted and analyzed in a large Finnish newspaper article data­ provements of subject access tools are primarily aimed at, pin­ base. The original queries were elicited from journalists who pointing the interplay between the different subcomponents of also evaluated the relevance of the search results. Each query any subject facility: a) the fe atures of the indexing languages was searched in three distinct modes : basic search, synonym used; b) the indexing principles used; c) the design and structur­ search and related tcrm search. The basic searches contained ing of the database; and d) the possibilities of the technical retrie­ only terms included in the journalists' original query statement. val facility, the search mode, and query languages. Thecontribu­ The logic of the basic search was developed by the researchers. tion summarizes the manifold interactions between the four sub­ In the synonym searches, the terms of the basic search were ex­ components listed. Any successful retrieval will heavily depend tended by disjunctions of the synonyms given by the searching on the design ofthese components considering their interactions. thesaurus without modifying the overall logic of the basic (Authors) search. In the related term searches, the disjunctions in the sy­ nonym searches were further extended by disjunctions of re1ated terms (quasi synonyms) given by the searching thesaurus again Furniss, P.: A proposed methodology for examining the provision without modifying the overall logic of the synonym search. The of subject access in the OPAC. searches in each mode were analyzed W.r.t. relative recall and Int.Classif. 17(1990)No.2, p.85-90, 22 refs. precision by setting the relative recall ofthe third mode searches to 100 %. The average relative recall in the first mode searches The development of the OPAC has stimulated renewed interest was ca. 45 % and in the second ca. 82 %. The average precision in the provision of subject access in the catalogue. Significantdif­ values were for the modes one, two and three were 51 %, 41 % ferences between the new and earlier formats have prompted the and 33 %, respectively. The substitution of synonyms shows a suggestion that there should be a radical reappraisal ofthe provi­ considcrable increase in recall with a negligible decline in preci­ sion made. A methodology for such a reappraisal is proposed, sion. based on the proposition that a radical reappraisal will need to (Author) refer back to the theory on which subject access is based. The Five Axiom Theory developed by R.Fugmann is outlined, which is considered to provide a possible basis for the reapprai­ sal. Some existing data derived from a cooperative cataloguing environment is described which is considered to be relevant to the axiom in Fugmann's theory which is concerned in part with Zeng, Lei: Establishing a compatible general the significance of consistency in indexing. The data is analysed vocabulary in China: the capability. in terms of consistency, with the objective of indicating what inLClassif. 17(l990)No.2, p.91-98, II refs. kind of results might be obtained from the application of the pro­ posed methodology. The conclusion drawn from the results of The study attempts to findan access to the establishment of com­ the analysis is that the methodology may well be worth further patibilityamongChinese thesauri, that is, to build a gcneral com­ exploration. Such exploration will need to examine amore extcn­ patible Chinese vocabulary based on existing thesauri in China sivedata sample than the limited one used for the analysis under­ and abroad. After a general analysis of the factors influencing taken in the article, and will need also to examine the role other compatibility between thesauri, the compatible capabilities be­ axioms of the theory might play in implementing a reappraisal tween special thesauri and general ones, as well as between of subject access. (Author) Chinese thesauri and English ones are discussed. (Author)

This contents page may be reproduced without charge INTERNATIONAL CLASSIFICATION IC

Scope Aims

The more scientific data are generated in the impetuous present Thus, INTERNATIONAL CLASSIFICATION is meant to be limes, the morc ordering energy needs to be expended to control a programme for the improvement of classification methods these data in a retrievable fashion. With theabundance afknowl­ and processes, a forum for discussion for all those interested in edge now available the questions of new solutions to the order­ the organization of knowledge on a universal or a subject�field ing problem and thus of improved classification systems, scale, using concept analytical and/or concept-synthetical ap­ methods and procedures have acquired unforeseen significance. proaches as well as numerical procedures and comprising also For many years now they have been in the focus of interest of in­ the intellectual and automatic compilation and use of classifica­ formation scientists the world over. tion systems and thesauri in all fields of knowledge, with special Until recently. the special literature relevant to classification attention being given to the problems of terminology. was published in piecemeal fashion, scattered over the llUll1Cr­ aus tc('hnical journals serving the experts of the various fields, INTERNATIONAL CLASSIFICATION publishes original such as articles, reports on conferences and similar communications, the Newsletters of FIDICR (Committee on Classification Re­ philosophy and science of scicnce search of the Intcrnational Federation for Documentation), science policy and science organisation and COCTA (Committee on Conceptual and Terminological mathematics, statistics, and computer science Analysis), book reviews, letters to the editor, and an extensive library and information science annotated bibliography of recent c1assirieation and indexing lit­ archivistics and museology erature, covering between 400 and 500 items in each issue. journalism and communicationscience industrial products and commodity science INTERNATIONAL CLASSIFICATION should therefore be terminology, lexicography and linguistics available at every university and research library of every country, a t every information cen ter, at colleges and schools of Ii­ brary and information science, in the hands of everYbody inter­ Beginning 1974, INTERNATIONAL CLASSIf'ICATION ested in the lields mentioned above and thus also at every office has been serving as a common platfonn for the discussion of for updatinginformation on any topic related to the problems of both theoretical background questions and practical applic

Instructions for Authors JVIll"llal Relerellr'es should contain the names and initials of all authors, full title of the publication, abbreviation of the journal according to the ISO Standard 4, volume number, year of pub­ Mal1l1.\·cripts may be submitted in either English, German or lication in brackets, issue number and first and last page num­ French to the editor-in-chief in two sets (the original and one bers. Monograph References should give the name(s) of the copy) typed in double space, comprising between 1500 to 3000 author(s), full title, edition, place of publication, publisher, year words. They should be accompanied by an English indicativeab­ and the number of pages. stract of I 00-200 words. The contributions arc refereed. Examples: Mater, E.: Human intelligence as a precondition for Criteria./bl" acceptance will be appropriateness to the field of the the machine processing of knowledge. Int.Clas­ journal (see Scope and Aims), taking into account the merit of siCi 5(1 988)No.3, p.125·132 the content and its presentation. Papers arc accepted with the un­ Hunter, E.J.: Classification made simple. Aldershot, GS: derstanding that they have not been published, submitted or ac­ Gower 1988. 115p. cepted for publication elsewhere and that, if the work received of­ JIIustrations should be restricted to the necessary minimum. ficial sponsorship, it has been duly released for publication. Graphs and diagrams should be supplied as black and white Authors will usually be notified within 6 to 10 weeks. Unless drawings suitable to reproduction. HalFtone illustrations specifically requested, manuscripts or illustrations will not he re­ should be sharp, well-contrasted glossy prints. Itlustrations turned. should be numbered lightly with a soft pencil on the back. Num� The text of the contributions should be structured by subhead­ bered legends should be attached on a separate sheet. ings; it should contain (a) an introduction, stating the pureose, Tables should be typed double-spaced on a separate sheet, con� (b) a description of materials and methods in sufficient detatl, (c) tain a number and a title at the top and be cited in the text. Each information on results or systems developed, and (d) a conclu­ column should have a heading. sion on and/or summarization. Authors shall receive 25 reprillls of their papers free of charge. References should be listed at the end of the paper with the num­ Additional reprints may be ordered. bers in brackets referring to such numbers in brackets within the Corrections in galley proof exceeding 10% of the typesetting text part. costs will be charged to the author. Additional notes should be indicated in the text by lifted single Contributions on micro flexible discs are welcome if delivered in numbers (e.g. text) and equally collected with their texts at the the ASCII code. Such contributions must be accompanied by end oflhe paper under the headingNotes. their printouts on paper. Editorial

Knowledge Organization and the OPAC More than 25 years have passed since the day we had from August 14-17, 1990, and particularly in the seminar the exciting, brand-new experience while attending the of Charles R.HILDRETH and Karen Markey annual convention of the then American Documenta­ DRABENSTOTT, this contribution will undoubtedly tion Institute: on a screen in Philadelphia, PA it was be a most welcome means to get suitably prepared! possible to retrieve literature from a rather remote place, And the summing up? Isn't it like this: the better we probably from Yorcktown Heights, NY - an incredible know our concepts and succeed in mastering them syste­ event at the time - 1964! matically, the better will we be able to gain control of Today, every scientist and scholar wishing to do so can both old and ncw knowledge through our Online Public have the literature he/she requires displayed on the own and Personal Access Catalogs (OPPACs). Therefore, thc PC and perhaps also -in a not too distant fu ture - on the knowledge of concepts - our units of knowledge - and own TV set, and in libraries the Online Public Access their systematic rcpresentability still strikes me as the Catalog (OPAC) has become and is increasingly becom­ very heart and soul of everything recognized and done in ing the fa vorite instrument (if not toy) for online retrieval classification,indexing and organization afknowledge. of available stocks, just as if this were the most natural This applies also, and particularly, to the topic of the thing in the world. Not only in the relevant literature can final article in this issue, on which we, too, wish to congra­ be browsed in this way - one would also like to get tulate its author, our little Chinese colleague Lei ZENG through to the knowledge contained in this literature. from Wuhan University in the People's Republic of But, as happens to be the case with nearly all things in China, soon to be a PhD from the University of Pitts­ this world: nothing is perfect, and all sorts of problems burgh, PA. This contribution, already well known still await a solution. Just how far have we got in securing among the circle of her colleagues at the School ofLi­ adequate and consistent access to the contents ofa publi­ brary and Information Science, was selected for the cation? How can it be realized? Will the use of a thesaurus "Orner Award" (an award established in honor of Ca­ in retrieval in nonindexed texts prove successful? This is therine Ofiesh Orner, a deceased distinguished alumna the question explored in this issue by Finnish authors of SLlS which is given annually for the best scholarly Jaana KRISTENSEN and Kalervo JARVELIN (in paper in information science). Lei Zcng sent her paper to "The effectiveness of a searching thesaurus in freetext us on a Macintosh floppy disc, which our PC unfortu­ searching in a full-text database"), if for the time being nately fo und to be non-compatible. However, a Berlin still on a 'laboratory scale'. Or is consistent indexing, company was able to write in the pure text via OCR and with or without a thesaurus and a classification system, send us a 'readable' disc. The tables and Chinese charac­ the key to success, as is examined by Peter FURNISS in ters, too, could be (separately) transferred, but they took his contribution "A proposed methodology for examin­ up so much memory space (4 megabytes!) that the trans­ ing the provision of subject access in the OPAC"? The mission task was too much for our system. (This just as a wealth oftopics concerned and the vast literature pertain­ little hint from experience to anyone interested in such de­ ing to them are what Winfried GODERT and Silke tails!). HORNY have chosen to systematically explore and pro­ We are happy that Lei Zeng tackled a subject which cess in a very detailed contribution ("The design of sub­ years ago was already treated in detail in IC but un­ ject access elements in Online Public Access Catalogs"), doubtely still contains many unanswered questions --par­ which we are happy to present to our readers in tlus issue ticularly regarding the aforementioned problem of the re­ in unabridged form, as translated from the German by quired intensive preoccupation with our concepts and Jacques ZWART. their necessary as well as facultative conceptual rela­ To those of us planning to participate in the firstinter­ tions. We hope for fu rther contributions in this direction! national ISKO Conference at Darmstadt, Germany, Ingctraut Dahlberg

Int. Classif. l7 (1990) No.2 Editorial 65 Winfried Godert FH BibUotheks- nDd Dokumentationswesen, Koln Silke Horny UniversiHit Konstanz, Stidwcstdelllscher Bibliolheksverbund

The Design of Subject Access Elements in Online Public Access Catalogs

G6dcrt, W., Horny, S.: The design of subject access clements in catalog users cannot estimate how much information Online Public Access Catalogs. they have failed to find. InLClassif.17(1990)No.2, p.66-76, 73 refs. As user problems in subject searches the fo llowing ca­ Experiences with retrieval in OPACs suggest thal there are tegories have crystallized out2: major problems for the users of such systems, such as choosing the 'correct' subject access vocabulary, narrowing or broade­ ning the set of items retrieved regarding the search interest. Such a) Difficulties in selecting the correct search te1'ms problems relate to two facls: (1) that in the early OPACs subject The user has difficulties in squaring his search subject access has not been considered seriously enough and (2) that the complexity of the topic has not been rcally recognized. Using a with the OPAC's indexing vocabulary. The vocabulary typology of users' questions it is demonstrated which require­ admitted for searches is regarded as too small, the predic­ ments musts be mct by a successful online subject access. Im­ tability and/or fidelity (12) ofthe terms to be used is insuf­ provements of subject access tools are primarily aimed at, pin­ pointing the interplay between the different sUbcomponents of ficient. Lacking information on the vocabulary to be any subject facility: a) the features of the indexing languages used and lacking conceptual links within the vocabulary used; b) the indexing principles used; c) the design and structur­ fr equently make the users select too special or too general ing ofthe database; and d) the possibilities ofthe technical retrie­ val fa cility, the search mode, and query languages. Thecontribu­ search tenTIs. These statements apply both to the use of tion summarizes themanifold interactions between the four sub­ verbal documentation languages and to that of classifica­ components listed. Any successful retrieval will heavily depend tion systems, for which the verbal class designations fre­ on the design ofthese components considering their interactions. (Authors) quently are not inverted along.

b) Difficulties of increasing the hit rates A large number of searches ends with the user obtain­ ing no or only a too small quantity of hits. The propor­ 1. Problems in Using tion of searches that end in failure lies between 35% and In the firstdevelopment stages of Online Public Access 57.5%. It may be assumed that the actual proportion is Catalogs the subject access element of this new medium even higher, as the number of hits does not guarantee was generally neglected, assuming as one did that this that the user has in fact found relevant information. A novel catalog form, like the traditional catalogs, would major part of the searches that have failed is due to the be mainly used as a findingtool for known items. user's unawareness that he must put in a search term However, user studies on Online Public Access Cata­ from a controlled vocabulary. Lack of knowledge about logs idFicated a shift in user behavior in that far more sub­ the indexing vocabulary thus fr equently makes for very ject searches were being carried out than had been antici­ low hit rates. pated. In the USA, numerous user studies have been car­ ried out on OPACs (1-9 and 72) which revealed a propor­ c) DifJiculties of reducing the hit rates tion of subject searches of an average 51 % (10). It also occurs very frequently that the user obtains very Noteworthy in this connection is the fact that this in­ high hit rates. Unaware as he is of existing limitation crease in subject searches has not gone hand in hand with possibilities, he may have to leaf in these cases through a corresponding increase in success in retrieval (9, p.153). more than 100 titles turned up (10, p.84). While in his search in the online catalog the user is far It is particularly these two latter points which need to more perseverantl than in the traditional catalog, he is be paid increased attention. In the initial stage one is in­ not in the same measure more successful (9, p.151). A clined, when using an OPAC, to rejoice over every hit query among OPAC users revealed that 85% of those which the qualitatively improved search possibilities, questioned are content with their searches. But an ana­ compared with the time-honored ones in traditional cata­ lysis of the query results reveals likewise that 56% find logs, enable one to achieve. However, as document nothing or only a part of what they are looking for. This stocks grow and familiarity with the system increases, result should, however, be viewed skeptically, as online one will feel an increasing desire for precise results wi th re-

66 Tnt. ClassiC 17 (1990) No.2 - GodertJHorny: Subject Access Elements in OPACs speet to the query and will note that the avoidance of b) the data records that become accessible through thc noise constitutes the major problem. For meeting the Online Public Access Catalog; aforementioned requirements, an optimized subject ac­ c) the users utilizing the Online Public Access Catalog cess element should provide tools which, in their results, for their search questions; neither increase noise nor entail information losses. d) program routines for facilitating such search ques­ tions. 2. Developments in Subject Access Elements Design If an online public access catalog is to offer possi­ The difficulties experienced by OPAC users may in bilities for subject searching it is necessary to take all part be explained by the significance - or the lack of it - at­ these factors into consideration in the planning and to tributed to the subject access elements in the first develop­ balance them among one another. mentstages of this medium. Therefore we wish to present With respect to the data records it must be clarified for here a brief overview of the OPAC's development his­ subject searching purposes what vocabulary will be tory. C.HILDRETH subdivides the existing Online Pu b­ made available for retrieval. This question concerns both lie Access Catalogs into three generations (13). the indexing method (extraction or addition 3 Most online catalogs of the first generation have come method) and the indexing principle (co-ordinate or syn­ 4 forth from libraries' lending administration or catalo­ tactic indexing) . guing systems onto which a so-called user interface was The retrieval language must permit the user to have ef­ superimposed. The catalogs of this generation permit fe ctive access to the indexing vocabulary, To solve this only precornbined searches, as they are customary in problem one may have recourse to experience gathered card catalogs. These Online Public Access Catalogs were with retrieval systems in literature data bases. However, mainly conceived for known item, at most title keyword in comparison with literature data bases an Online Pub­ searches. In connection with the systems from which they lic Access Catalog has relatively little indexing vocabu­ came forth there are hardly any subject description data lary at its disposal (e.g. no abstracts). available. The dialogs are either strictly menu-controlled In addition, the documents contained in the online or based on a command language. public access catalog are highly multidisciplinary. Thus, It was only in catalogs of the second generation that in retrieval in the online catalog more attention needs to subject searches via keywords and subject headings be­ be paid to vocabulary selection and design and control. came possible. At the same time these online catalogs The complexity of the subject is increased by the fact have a sophisticated retrieval language at their disposal that special requirements are imposed on the user inter­ which uses, among other things, Boolean operators. Fur­ face of an online catalog. Library patrons are not trained ther aids in searching are browsing in the index and the information searchers, and the user interface must be so possibility of systematic searching for call numbers or no­ designed that subject searches can be performed even by tations. Second-generation OPACs have a more flexibly inexperienced users. Therefore, in connection with the designed user interface, with a shift from menu to com­ user interface, consideration must be given to whether mand mode and vice versa frequently being possible. and how it can assist the user in his subject searching, but This differentiation between inexperienced and experi­ also to when it limits subject searching possibilities, enced users should be overcome in the OPACs of the In addition, thought must be given to whether pro­ third generation. In these systems the user interface is gram routines can be made use of for assisting in subject more adapted to the users' needs (e.g. input of questions searching. As the final important factor, the layout, too, in natural language). At the same time these catalogs pres­ plays a part. The user should be able to recognize from ent improved possibilities for subject searching. Frce­ the display of the subject information whether the docu­ text and controlled vocabularies are mutually inter­ ment is of interest to him. The subject information must linked and further subject access possibilities arc inte­ therefore be clearly formulated and plainly put in relief. grated into the data records. These systems also have cor­ rection algorithms at their disposal which try e.g. to cor­ 4. Typology of User Questions rect any textual elTors or to perform an evaluation of the To be able to lay down what a subject access element retrieval results. Online Public Access Catalogs of this should contain, we will attempt to determine on the basis generation are directly developed systems rather than of a typology of user questions what different require­ having been fitted retroactively to a cataloguing or lend­ ments occur. On the user's actual subject information ing system. needs only little is known. The enquiries observed in the analysis of transaction logs cannot be made use ofin this 3. Subject Access Elements form, for it must be assumed that the users tailor the for­ Before discussing the various parls of the subject ac­ mulation of their questions to the search possibilities cess elements it needs to be clarified just what is meant by available. From such a typology conclusions should be the subject access clement and what requirements can be drawn as to the design of and the possibilities to be of­ imposed on it. fered by the retrieval components. The subject access clement is a highly complex entity having several components. Carol A.MANDEL de­ a) Questions enquiring after simple subjects (lexicon ques­ scribes it as a three-dimensional puzzle in which four mu­ tiolls) tually independent factors meet (14): Such questions can - as a rule - be represented by a a) The design ofthe user interface; word or by annotation of the indexing language used in

Int. Classif. 17 (1990) No.2 - G6dertfHorny: Subject Access Elements in OPACs 67 the system, i.e. by the semantic component. For answer­ As already indicated before, this case is particularly fr e­ ing such questions, therefore, the availability of the voca­ quently mentioned in the user observations as most sys­ bulary together with its relational structures would be tems so far do not have sufficient possibilities at their dis­ sufficient. However, as an actual user's need (rather than posal for a suitable reaction. In future design, this point as a query actually observable so far) such questions are should, above all, be given special attention. As we will far rarer than usually assumed (10, p.62). see later, this is also the point where one should not con­ fine oneself to using only a single type of indexing lan­ b) Questions relating simple subjects to fo rmal data guage, but rather honor the long-proven experience that (auth01� publisher, year o.f publication , etc.) a combination of several description methods (verbal This is realized by pre-set or fr ee linkages (possibly and classificatory) promises the best results. using Boolean operators) of the categories concerned. Al­ though a typical characteristic of online catalogs, this is 5. Possible Components of the Subject Access Element not originally a problem of subject access element design. 5.1 Indexing Language The first sUbcomponent to be considered is the index­ c)Questions relating simple slll�iects to data pertaining to ing language used. Already on this level there is a great time and/or space diversity. As a first differentiation, the traditional divi­ This type of questions might well occur quite fr e­ sion into classification systems and verbal indexing lan­ quently: at least in the free combinability of the various guages is useful: for further-going considerations a finer aspects, it surpasses the possibilities of any conventional differentiation is required. catalog. Realization requires the availability of corre­ Because of their versatile possibilities of use in index­ sponding indexed items (subject headings, notations, ing and retrieval in online public access catalogs, classifi­ etc.) as well as a retrieval component permitting the com­ cation systems are once again being paid increasing atten­ bination of these items. Such combining can take place tions. As numerous contributions have already pointed through the linkage of corresponding index categories or out the usefulness of classification systems in online re­ through the free application of Boolean operators trieval, suffice it here to recapitulate only their chief ad­ equipped in a wide variety of ways, not excluding the com­ vantages. In online retrieval, classification systems can in plex bracket logic as we know it from the commercial re­ a variety of ways be successfully applied to reduce the trieval languages. noise rate in retrieval and to improve the precision of the A voluntary fo regoing of such search possibilities in results. The fo llowing possibilities of use exist: an online catalog could hardly be said to measure up fully In the first place the possibility of narrowing or expan� to the expectations that will be placed in these systems, at ding concept fields. This possibility is only offered, how­ least in the more progressive ones among them. ever, by classification systems with expressive (structure reproducing) notation systems6. Thus, such classifica­ d) Questions giving expression to complex, a jJosterr'orire­ tion systems also help to narrow or expand hit rates. Fur­ lations (including directional relations resulting e.g. Ji'om thennore the use of classification systems provides a actions) possibility to place verbal expressions into a conceptual It is probably this type of question which gives rise to framework and thus to avoid ambiguities. Particularly in the majority of controversies as to the necessity ofrealiz­ multidisciplinary data bases verbal search terms may be ation in library catalogs. Reasons for giving this type of ambiguous. By linking the search term with a class from questions a certain amount of attention are presented by the classificationsystem conceptual unambiguity may be the fact that tills type is anything but rare, while on the realized (22). With the aid of classification systems it is other hand - particularly in multidisciplinary and grow­ also possible to specifically exclude classes of subjects ing document stocks - the answers to these questions are from the search. Moreover, working with a classification particularly unsatisfactory if the proper searching tools system fa cilitates searches for thematic subjects that can are not available. Not in the last place, approaches to­ only be formulated with difficulty by verbal search terms ward a solution of this problem can already be found in ·(23). If it is displayed in what contexts the unconerete the great majority of indexing languages - even jf usually words occur in the classification system the user can spec­ only (which admittedly is a serious setback to post�coor­ ify his question fu rther. dination in online catalogs) in firmly joined precombina­ The use of a classification system also permits thc dis­ tions in the vocabulary, whichthus are not available for play, during a search, of systematically ordered title lists free combinability (15). fu rnishing a bettcr overview in the case oflarge hit rates. Realization requires in any event the use of Boolean Of interest, finally, is the question whether access to operators. For directional relations, syntactic aids must the classes is possible only through input ofthe notations be made available which, through suitable representa­ or via the class designations. If maximum comfort of use tions, must also become visible on the surface of the in­ is striven for, only the latter possibility would probably dexing language. Whether these syntactic aids are also to be acceptable. But to what extent a verbal access may be be used directly for retrieval purposes is a question requir­ an alternative to a verbal access component is a question ing deeper consideration. that is not easy to answer and depends in large measure on the details of concretization. The mere inverting of e) Questions in the nature of cases c) 01' d) above that are class designations assuredly will not produce a vocabu­ related to or circumscribed by means oj/onual data lary suited to serve as a verbal access possibility. The

68 Int. Classir. 17 (1990) No.2 - G6dcrtJHorny: Subject Access Elements in OPACs class designations would at the very least have to be prow represented by a single descriptor (precombination), or cessed for serving as a retrieval tool. Measures for termiw should it be separated into its parts and be represented by nology control would have to be taken, descriptors for inw several descriptors in indexing (postcoordination)? Both 9 dividual concepts be admitted and the vocabulary be enw procedures have their own advantages and setbacks , so riched by fu rther search terms (I8). Nevertheless, class that in selecting the descriptors a middle path should be designations are increasingly regarded as a good alternaw chosen. Thus the German standard DIN 31623 Part 2 7 tive to verbal indexing languages . Regardless of how the (28), for example, presents procedures for concept disas­ verbal access vocabulary is designed, a linkage with the sembly and fo r concept assembly. It should be noted here notation system should be guaranteed in any event. that the designations are not broken up into their mor­ Verbal indexing languages constitute, at first glance, phological components. Since the a priori relations must the more homogeneous world when compared with be reproduced in the vocabulary, a semantic resolution classification systems. But at the very least one should of concepts should be performed. determine whether one wants to work with individual In the retrieval of complex subject headings, coordi­ subject headings, with thesaurus descriptors or with prew nated indexed items fr equently entail disadvantages. combined subject heading strings, whether and to what Linkage of individual subject hadings, which always rcp­ extent one wishes to engage in terminology control and, resent individual concepts exclusively, is to permit, in if so, how the vocabulary is to be attended to in so doing. postcoordination, the reproduction of complex topics, Here it is imperative that it should be recorded, and be which as a rule can only be imperfectly achieved through kept available at all times in a suitable instrument, what such linkage using Boolean operators (see ch.5.4). Coor­ vocabulary has already been used w all this without over­ dinated indexed items are only insufficientlysuited to rep­ looking the fact that not every type of vocabulary is resent complex SUbjects. suited fo r every type of retrieval. This depends in large measure on the indexing principle. The limits of the possi­ b) Syntactic indexing bility ofretrieval by verbal means are reached at the latest Through syntactic indexing, more precise results are when hit rates need to be increased or decreased. On the to be achieved in retrieval and noise thus to be avoided. vocabulary level it is not sufficiently possible to carry out Presence of syntax and/or use of syntactic aids is distin­ concept limitations or expansions. For practical usc, a guished on three levels: floating transition from a verbal indexing language to a A: The level of the indexing language vocabulary avail­ classification system would be particularly helpful. The able for indexing (precombined syntax). user could then start his search on the verbal level and for B: The level of the indexed items or that of the indexer's concretizing his question change over to the classifica­ as user of an indexing language, i.e.: can the indexer tion system level. synthesize complex propositions from components of the vocabulary using syntactic aids? 5.2 Indexing principle used c: The level of the user of a documentation language, At this point it is necessary to give attention to the in­ Le.: can the user address to the documentation sys­ dexing principle, independently of the indexing lan­ tem, in synthesized fa shion, complex questions made guage. Because of the technical possibilities of postcoor­ up of components of the vocabulary, using syntactic dination through Boolean operators, access is frequently aids'? had in the OPAC to all data available, regardless of To represent a posteriori relations on the vocabulary whether they are suited to such access or not. This can be level, precombined descriptors must be formed, a proce­ explained by the wish not to forego under any circum­ dure we will call 'prccombined syntax'. Now, while these stances the possibility of postcoordinatingly relating precombinations have a high degree of precision, the re­ search terms to one another. This, however, will by no sulting volume of the vocabulary renders the predicta­ means necessarily improve retrieval results in every bility of the descriptors more difficult. Also, prccom­ single case; on the contrary, precombinations strikingly bined concept links render links on levels B and C proble­ and extensively present in the vocabulary - in particular matical. On the vocabulary level, therefore, it is first and hidden syntactic interrelationships in composites - may foremost the semantic (a priori) relations which should quite negatively affect the predictability and fidelity of be uncovered. 8 the indexing vocabulary used . While the semantic concept links are recorded in the vow Coordinate and syntactic indexing will at this point be cabulary, the expression of syntactic a posteriori rela­ brieflyintroduced and be examined as to their applica­ tions on the indexer level requires the availability of suit­ tion possibilities in retrieval. able tools. To this end, various possibilities suggest them­ selves (29, 30). The use of prepositions between the sub­ a) Coordinate indexing ject headings and the fo rmation of descriptor strings as In coordinate indexing the descriptors are lined up well as the form ation of subsets (using links) and suitable alongside each other irrespective of any relational view­ weighting indicate that a relationship exists among the points whatsoever. This indexing principle was de­ descriptors. A directional relation can, however, only be veloped mainly for postcoordinate retrieval. A major made explicitly clear through the assignment of role oper­ problem in this connection is the question to wha t extent ators. complex concepts should be broken up. Should a com­ If, for postcoordination purposes, one wishes to use plex concept consisting of several individual concepts be components of precombined indexed items or of indexed

fnt. Classif. 17 (1990) No.2 -- Godert/Horny: Subject Access Elements in OPACs 69 l2 items structured according to the syntactic indexing prin­ vocabulary to always produce noise as welI , this ciple, problems arise that need to be solved (see IS): method will likewise result in an increase of the tlOice a) Are the rules leading to the formation of the precom­ rate. The subject search should therefore be performed bined expressions of such a nature that the individual firstof all in the controlled vocabulary, whose processing components have expressive force with a view to post­ and/or indexing for retrieval purposes should be im­ coordinate searches? In other words, do the various proved. The use of natural language for fr ee-text sear­ components have a complete semantic identity or ches recommends itself, on the other hand, for searches 13 does their complete understanding rather require a se­ for individual concepts and highly new terms . In the re­ mantic information transfer fr om other parts of the lated indexing one should not forget to provide, besides complex of precombined or syntactic indexed items? access to the individual concept, also the possibility of ac­ b) How are the precombined or syntactic indexed items cess to an abstract superordinated concept. Only in that to be processed to make them suitable forpostcoordi­ way can the completeness ofthe subject search be guaran­ nate searches? In other words, what measures does teed. one take for keeping intact or restoring necessary se­ mantic units? A search through the totality of the pre­ In addition to the fieldsfor the indexed items one may combined indexed items would certainly be an impo­ provide further fields for formal criteria, such as e.g. for sition on the user; in that case the OPAC would limit the form or language of publication (38). These criteria the search even more than ,a card cataloglO. If one can be combined with the search concepts and thus pro­ breaks up the precombined structures into individual vide fu rther possibilities for limiting the search. descriptors one will face the same retrieval problems as in the retrieval of individual descriptors (see 5.4 below), and the efforts invested in the formation of 5.4 Retrieval language the precombined structure are lost. A practicable sol­ The retrieval language on which one whishes to base ution seems to be the approach used e.g. by the Brit­ the Online Public Access Catalog must be selected in ac­ ish Library in its searches for PRECIS strings: After cordance with the indexing principle. This is particularly the input of one or two search words all strings are dis­ important if the documentation system features syntac­ played in which the search words occur. Thus the user tic indexing. The retrieval software must permit the user can draw conclusions from the precombinations dis­ to ask the documentation system complex questions con­ played as to theeontents ofthe documents and will ob­ sisting of components of the vocabulary while using syn­ tain more precise retrieval results (32), see also 00- tactic aids. But in the case of coordinate indexing, too, dert in (15). aids must be offered to iron out the setbacks resulting c) How are the implicit or explicit syntactic aids, if any, from the indexing principle. made utilizable for retrieval purposes? Answering these questions probably constitutes the a) Boolean operators most difficult problem to be solved in the design of sub­ The Boolean operators AND, OR. NOT are among ject access elements of online catalogs. In considering the the most important components of the retrieval lan­ indexing principle it undoubtely remains valid that syn­ guage. They are indispensable for performing postcoor­ tactic indexing cannot be foregone if effective searches dina ted searches, in which individual descriptors are to for complex subjects are to be rendered possible. be related to each other by linkage (39, p.532). Now while Boolean operators are a powerful retrieval instrument, 5.3 Search field access design their use is not wholly without problems, for it also fre­ The data of any bibliographic item are stored by fields. quently results in less than precise searching results. In processing them for search purposes it must be de­ Some ofthese setbacks will now be briefly presented here. cided whether specific access to individual fields is Searches with the aid of Boolean operators will also possible or whether several fields should be combined turn up titles that are more specific than the search query into one search field. as originally formulated. With the usc of Boolean oper­ Access to individual fields requires the user perfor­ ators only it is not possible to search exclusively for the ming the search to have exact knowledge of the structure search words desired. Documents indexed with more de­ of the data record. This approach also permits, however, scriptors than only the search words put in will likewise precise access to the controlled vocabulary. be reported as hits. Retrieval tests concerning recall rates have shown, however, that searches using controlled vocabularies ex­ Example: A search for "woman AND age" clusively do not produce the best retrieval results (33) could also turn up, as hits, the fo llowing titles (e.g., in­ (73). Such searches must, therefore, either be substan­ dexed with the following RSWK strings (RSWK � Rules tially improved or be supplemented by searches using for Subject Cataloguing)) other elements. As such, free-text clements have usually Woman I age I personal report been employed so far (34). Woman I age I fitness I guide In many systems, specific access to individual fields is Woman I agel living conditions I Germany Woman / age / mental hospital / narrative interview replaced by the fo rmation of a basic index 11 in whieh free-text and controlled vocabularies are offered jointly. Furthermore there is no guarantee that in the hits Since data inhomogeneity causes a search using free-text turned up the descriptors as put in occur in a conceptual

70 Int. Classif. 17 (1990) No.2 -- Gndert/Horny: Subject Access Elements in OPACs contest. They may also occur in the document wholly in­ the character sequence put in. Frequently this leads, how­ dependent of each other if two or more strings are as­ ever, to numerous undesirable words, as the given con­ signed to the document. ceptual realm is abandoned. One can try to mitigate this problem by employing limited masking, consisting inlay­ Example: The title: "Gervasi, Tom: Moscow's Supre­ ing down the number of characters to follow the charac­ macy" was assigned the following descriptors: ter sequence put in (42, p.120-127). USA / Arms race / Soviet Union In addition to right truncation, left truncation would USA / Strategic defense initiative also be desirable, permitting as it does the findingof com­ This title also turned up, however, in a search for: pound words of which the word put in forms the second Soviet Union AND Strategic defense initiative component. As compound words occur frequently, par­ ticularly in the German language, this element of the re­ Finally, it is not possible to express directional a poste­ trieval language would be a useful instrument (44). riori relations by means of Boolean operators. A final masking possibility would be masking the center of the word. This could be used in the case of devia­ Example: A search for "Evaluation of students by ting spelling or ifthe user is not sure ofthe correct spelling teachers" would be expressed by Boolean operators as: of the word. evaluation AND students AND teachers. Titles such as "Evaluation of teachers by students" d) Menu guidance would then also turn up as retrieval results. For the user to be able to employ the aforementioned retrieval possibilities for conducting effective searches, Particularly when employing syntactic indexing this he must have exact knowledge oftheir use and their effec­ would lead to unsatisfactory retrieval results (13). These tiveness. But as it cannot be assumed that all users are problems occur to an increased extent if one is searching able or willing to go into these retrieval intricacies, the on­ in a "basic index". line catalog should also put simplified retrieval possi­ Besides for searching within a search category, bilities at their disposal. Boolean operators may also be used for linking up For this purpose a user interface is created in which various categories among one another. To this end it is suitable presettings are performed which lay down necessary to put in, besides the search terms, also the rele­ search categories and perform automatic linkages as well vant category designations (38, p.438). For subject sear­ as, possibly, automatic maskings. ches this possibility is of interest if, e.g., subjects are to be Through menu guidance and 'suggestive prompts.J4 limited by formal criteria. the user is enabled to ask the system more complex ques­ The use of Boolean operators in Online Public Access tions. Though flexibility is reduced, the inexperienced Catalogs is not uncontroversial. It is fe ared that the user user still obtains satisfactory retrieval results. may fail to grasp the logic behind these operators, result­ ing in their incorrect use (40). Yet it is not possible to dis­ pense with these powerful linkage possibilities. Their use e) Numeric area queries is necessary for handling complex search queries in the A further important property of the retrieval language online catalog. is the scanning of numeric areas. In this way the user can Some of the problems occurring in searches with be enabled to set limits to his search questions by means Boolean operators can be avoided by using syntactic of numeric criteria (.e.g. publication year). The search aids, such as coupling indicators, on the indexer level (41). for numeric data constitutes a special problem. It in­ cludes e.g. in our aforementioned typology of user en­ b) Adjacency and context operators quiry the setting of time limits.lf, in addition to and be­ To alleviate the shortcomings occurring in searches yond a search for individual years, one also wishes to using Boolean operators, the user may be offered syntac­ query according to time intervals, it is necessary that tic aids. By means of adjacency or context operators it these data be stored in specificnumeric data ficlds so that may be indicated in what order, how closely together or access to them may be had using numeric relational oper­ in what data record the words sought are to occur, it ators (e.g. ). This is not in the last place a question of data­ being assumed in this connection that a relation might bank structure; and the existence of such data is a ques­ exist between these descriptors if they occur closely tion of the indexing language and the indexing conven­ together (42, 43). tions agreed upon. Adjacency operators may fu rthermore be used in sear­ ches in controlled vocabularies for laying down the order ofthe descriptors or for expressing the distance that is per­ 5.5 Search for indexed items mitted to occur between the descriptors in a string. Here it is necessary to distinguish bctween 'direct searching' and 'browsing'. No OPAC should be allowed c) Masking (Truncation) to offer just one of these two searching possibilities exclu­ For searches in the free-text vocabulary the retrieval sively. Both procedures have their advantages and set­ language must pennit masking to that morphologically backs. It should be possible to select between the two as related words may be searched fo r. In the case of right suggested by the nature and complexity of the query and truncation, all words are searched for which begin with according to the user's knowledge level (46).

Tnt. Classif. l7 (1990) No.2 - Godert/Horny: Subject Access Elements in OPACs 71 a) Db'ecl searching That such a procedure may be important is indicated A retrieval process is termed a 'direct search' if hits are also by a behavior study which showed that the user im­ reported immediately upon the input of the search mediately stops reading the list when he has found a rele­ term(s). This approach thus leads directly to a title dis­ vant word, even if further relevant words are contained play without any 'detouring' through any indexes. Com­ in subsequent parts ofthe display (52, 53). pared with browsing, direct searching is much fa ster. The indexed items can be arranged either alphabeti­ However, it is not very error-tolerant. Search terms must cally or systematically. Alphabetic descriptor lists be put in in the correct order, i.e. according to the syntax should indicate synonyms and inform the user of the of the retrieval language, and in the correct spelling. In preferred term through appropriate cross-references. searching for the correct search term, a direct search In an expanded form they can also indicate hierarchi­ would be the simplest solution, but for the user the most cal relations (super and subordinate concepts) as well as unsatisfactory one. After having put in a search term, the associative relations. The user's attention can be called to user is informed in reply whether this term is stored in the these relations during his search by socalled 'action catalog memory or not. In the latter case he must put in a codes', which make it easier for him to handle the vocabu­ new word. The direct approach is useful only if the term lary lists and which help him in his search (54). In the case put in matches a descriptor or is marked as the synonym of subject heading strings, Keyword-in-Context or Key­ of a descriptor. Iftheword put in is not a descriptor, atten­ word-out-of-Context lists should be made available to tion should be called to the preferred term (45, p.12!). A permit access to subheadings as well (4, p.50-51). linkage fr om the non-descriptor to the descriptor should Alphabetic lists furnish a rapid overview ofthe vocabu­ not take place without the user being so informed. The lary and help in the correct formulation of the search user would only be needlessly confused ifthe retrieval re­ word (53), thus being able to provide effective help in sear­ suit displayed uses a different term from the one he has ches for simple subjects. They have the disadvantage, put in. Hence the system should call attention to the however, that they can only illuminate relations between preferred term and, after confirmation by the user, con­ subject terms resulting fr om the alphabetic sequence. tinue the search with the descriptor. A renewed input by Systematic descriptor lists are particularly useful for the user should not be necessary. Direct searching determining the hierarchical conceptual level of the word becomes problematical if the search produces no results sought, as well as for correction purposes if the subject although the user has made only a syntax mistake. These search produced unsatisfactory results. errors may be straightened outin part by improvement al­ For the representation of hierarchical conceptual gorithms. structures several models are available: The hierarchical conceptual relations between the b) BrolVsing subject headings are displayed in the form of search The search process is called browsing if title displays trees which, in thesaurus-like manner, indicate the are preceded by a search in the index. This search can, va­ super- and subordinate concepts (54). rying with the search process, be conducted in variously Through the input of a notation, browsing through ordered descriptor lists or in abbreviated-title displays. the class designations of the classification system In these indexes the user can also look at preceding or sub­ becomes possible (20). sequent entries. 'Brbwsing' thus describes a search pro­ Likewise available is the possibility to systcmatically cess taking place in like manner as a search in a card cata­ search for documents. 'Shelf-list browsing' simulates a log (48). systematic search at the shelf. The user puts in a call num­ Browsing through the search vocabulary is generally ber, upon which a systematically orderd title list is dis­ found useful by users trying to select the proper search played. Such a search prcsupposes, however, that the call term (49). This can undoubtedly be explained by thc fact number is known to the user. that browsing in the OPAC is patterned after browsing in the traditional catalog. 5.6 Display of indexed items of documents found Browsing is slower than a direct search. The user must start by putting in a search word and then - after having The display of the indexed items in the data record of browsed through the index - again select a search word. the documents found can likewise be offered as an aid in On the other hand, this browsing furnishes him an over­ subject searches. view of the indexing vocabulary, which can then be of as­ Having found relevant information, the user has the sistance to him in fu rther searches (46, p.64). data record displayed. In reading the data rccord the user For browsing in the descriptor lists the user puts in a estimates the subject relevance of the title. If the indexed search word. This term is either complete or is masked items are displayed, too, he may obtain the helpful infor­ (truncated) by him15. In some systems the search word is mation under what descriptors or notations he may ob­ masked also implicitly, in which case the system guides tain even morc information (10, p.IIS). 16 the user to a specific point in the index fr om which he Display design varies widely . To fa cilitate the search should be able to browse both forward and backward. for further subject information the aforementioned data If long lists are displayed, browsing proves to be may be put in relief, e.g. by doubling their luminosity (10, problematical, as the user easily loses patience and p.ll5). merely skims over the contents of the screen. In these In displaying notations the class designation should in cases hierarchically graded lists as introduced by any event be displayed alongside so as to facilitate the M.MASSICOTTE are recommendable (51). user's understanding.

72 lnt. Classif. 17 (\990) No.2 - ODdertfHorny: Subject Access Elements in OPACs 5.7 User interface users can rank the importance of the various search The user interface constitutes the connecting link be­ terms, upon which the computer recalculates the relev­ tween man and machine. Particularly because of the dif­ ance of the various documents. ferences between user groups, user interface design is a The Harper algorithm in the OKAPI system works in difficult undertaking, since every library patron should much the same way (63). If, in OKAPI, several search be able to use the online catalog without being familiar terms are put in, they are first linked together by an im­ with it in detail. plicit AND. If the search yields no hits, the socalled Varying with thcir mentality, their fa miliarity with the HYPER-OR is put in which is to determine those docu­ system and their experience, users will impose different ments which can roughly satisfy thc query. The selection � requirements on the online catalog. While the inexperi­ of these approximate hits is based on H rper's algo­ enced user will let menus guide him through his search, rithm, whose basic idea is that the search terms grow in the more advanced user can ask questions of greater com­ importance as they become more specific. The less fre­ plexity. The user interface should therefore offer various quently a search term occurs in the database, the more im­ levels of difficulty so as to satisfy the different needs. A portant it will be. In selecting the titles, those ones will be change-over at any desired stage should be possible17. considered first which contain as many as possible of the One should beware of letting the inexperienced user's rare words. needs be the decisive factor in user interface develop­ Since all these methods influence ony the recall but not ment. The users must also be given the possibility of self­ the precision ofthe result they cannot be regarded as a sol­ development in their searches. Systems designed exclu­ ution to all problems brought to light by the questioning sively for user-friendliness usually offer only a limited of users. measure of retrieval conveniencelS". Of great assistance in retrieval are the 'help' functions, b) Expansion oI the search vocabulary especially the 'suggestive prompts' which at various From various quarters it has been suggested to enrich points in the search make suggestions to the user on how the data records by information from tables of contents to continue his search. So that they may be properly and indexes. In that way the user would have more voca­ understood and utilized by the users, these help functions bulary at his disposal for keyword searching, which should be very carefully formulated (52). would increase the probabilitry of hits (64). In large data quantities this procedure will undoubtely create prob­ 5.8 Proposals for improvement lems resulting from lacking terminology control (65). The user studies mentioned in the beginning have Similar problems are to be expected if the various data given rise to various proposals for improvements, with records of a catalog are filled with data from various sub­ one school of thought trying to improve retrieval by stat­ ject description systems a trend which seems to be present istical and technical means, while another one advocates in West Germany (66-68). Through taking over data measures for improving the vocabulary used in indexing, from extraneous sources, this procedure can be realized the indexing principle applied, and database structure. in a relatively fa st and easy manner. The number of terms characterizing the contents of the document will in this a) Improvement algorithms way indeed increase, but because of the inhomogeneity Improvement algorithms are internal computer pro­ of the data noise-free retrieval will not be possible (69). grams to be used when a search has produced no results. Thus it does not appear to make much sense to stimulate Since many negative retrieval results are caused by mis­ the search for free-text elements by clogging these fields spelling of the search word, one can try to make up for with data. this error by spelling correction programs (58). The use­ fu lness of this technique is limited, however. c) Entry vocabulary More sophisticated algorithms look for words whose A helpful tool in subject access would be the formation pronunciation sounds much the same as the search word. of an entry vocabulary as proposed by M.BATES. The The Soundex algorithm is an examplc (59). aforementioned user studies show that users have diffi­ In the next step one can try to form morphologically re­ culty in findingthe correct search term. M.Bates explains lated words. These procedures include the automatic this phenomenon with the versatility of natural lan­ masking of the search terms and the reduction of the guage. As complete indexing consistency cannot be word to its stem, the so-called 'stelmning' (60). It is as­ achieved in any indexing system, the user should not be sumed here that a morphological relationship also im­ expected to always correctly determine the descriptor plies a semantic relationship, ej. that the newly acquired (54, p.361). In addition there is the problem that users search terms will be related contents-wise to the search preferably use their own search words in their searches term put in. without obtaining information on the controlled vocabu­ Other algorithms are used in efforts to expand the lary (l0, p.70). given semantic area. They try to find similar concepts to One solution to this problcm would be the develop­ the search concept. These techniques include the closest ment of two vocabularies: one used by the indexer and match search and the fuzzy-set approach (35), (61). one placed at the user's disposal. This 'end user the­ In addition, programs have been developed which per­ sam·us' would then serve as entry vocabulary. It would form an evaluation of the search vocabulary Cranking') contain numerous terms fr om the natural language and (62). In so-called 'user feedback routines' (10, p.1l7) the thus be far more voluminous than the controlled vocabu-

lnt. Classif. 17 (1990) No.2 -G6dert/Horny: Subject Access Elements in OPACs 73 lary. Thus the user might formulate his search term in the the user can correct his search question if the search term natural language without acquainting himself prior to selected by him turns up too many or too few documents. his search with the controlled vocabulary. Besides con­ Also, he can assess the relevance of the indexed docu­ ceptual quasi-synonyms this vocabulary would also ments already in the search state without first having the cover alternative spellings. titles displayed to him. The vocabulary is displayed The terms of the entry vocabulary would be connected either alphabetically or systematically, and the change­ with the controlled vocabulary by cross-references. over between both modes must be readily possible (e.g. Upon input of a search term, the controlled vocabulary by pressing a fu nction key). would guide the user to the computer's internal cross-ref­ The user interface should be so designed as to help the erence network (34), (54). user to develop his best search strategy. This includes in An 'end user thesaurus' of this nature would un­ particular that tools must be offered for selecting the doubtely be an effective tool, but also a laborious one, for search terms and for expanding or reducing the retrieval it needs continuous vocabulary control and updating. results. As the menu mode can offer only a limited Entry vocabularies need to be drawn up by each library measure of retrieval convenience, the user should be individually so that no blind cross references will occur. guided by these aids to the cOlmuandlevel. The user should only be guided to such descriptors as In conclusion we will quote Pauline A.COCHRANE, correspond to available literature. In addition, such an who has very ably summed up the entire discussion entry vocabulary needs to be continuously expanded and around the subject access element as follows: "The ques­ updated: The compilation of a further file comprising all tion is not free text versus controlled searching, nor is it those designations which did not lead to a descriptor Boolean versus non-Boolean searches. The question is would permit a continuous expansion of the entry voca­ 'How can we create a catalog that brings works together, bulary (19). docs not separate related subjects or conceal informa­ tion, and allows the user to search with ease andlittlediffi­ 6. A summing-up culty no matter whether the query is specific or general'" (71). Now what conclusions follow from the above? Here we will smmnarize a few criteria which in the authors' opi­ nion add up to a good subject access element. For subject searches in the OPAC the document must Notes: be assigned subject description data, as the free-text el­ It was found that 70% of the subject searches in traditional ements available in the data record are suitable only to a catalogs search for only one term (see (I I» . Tn an OPAC the limited extent for retrieval purposes. They should be sup­ user, however, searches on the average by 5 different terms (see 10). This can be explained probably by the fact that a fur­ plemented both by vcrbal and classificatory retrieval el­ ther attempt does not mean at the same time a change ofloca­ ements. tion. In addition the new medium challenges the user and en­ tices his joy to play with the computer (see 10, p.2) The controlled vocabulary must indicate the occur­ 2 See (9, p.124) and (3). ring conceptual relations. These include cross-references 3 The authors assume that an effective search in an OPAC to super- and subordinate concepts as well as to selected must be based on a controlled vocabulary (which means the application of the addition method). A mere free text search associative concepts. In addition to the controlled voca­ is deemed insufficient, as by the missing terminology con­ bulary. Further contents-characterizing data such as trol the retrieval result would show too high a noise rate. form headings, timeindications,etc., which through link­ 4 See Section 5.2 5 This is mirrored in the numerous references having ap­ ages with the vocabulary permit concretization of the peared on this subject, see (16-24). A negative assessment of search question, should be stored in separate fields. To classification systems in OPACs is given by 1.HILL in (24). permit searching for complex subjects the controlled vo­ 6 The present projects preparing the Library of Congress Classification for online retrieval show, however, that it is cabulary is based on the syntactic indexing principle. possible to attach to every classification system (also a poste­ Besides a verbal documentation language a classifica­ riori) a structure reproducing notation, see (25) and (26) tion system with structure-reproducing notation is also 7 Thus one can access lhepreparcd subject index aHhe Univer­ sal Decimal Classification in the online catalog ETHICS at available. This classification system features verbal ac­ the Technical University of Zurich (sec (27). K.MARKEY cess, i.e. transition from the verbal indexing language is also tried to search with prepared class descriptions in a pro­ possible. The classification system is an important tool ject utilizing the Dewey Decimal Classification, see (20, 21), cf. also (16). for improving the completeness and precision of recall. 8 A problem already referred to by R.FUGMANN, see (12). The combination of verbal and classificatory search el­ 9 See the detailed presentation in (12). ements is to offer the user a measure of searching conveni­ to That this request is not self-evident is to be seen, for in­ stance, in the utterance of E.FA YEN on "phrase search­ ence which leads to satisfactory retrieval results. The ing": "The process is analogous to what be required in the noise rate is to beas low as possible. It is only ifno results card catalog and does not present a problem for many sear­ are obtained on this basis that one should start searching ches." (31, p.76). II A 'basic index' is a multidisciplinary main index in which a in the free-text vocabulary. Improvement algorithms, number of indexes have been compiled into an additional too, should only be put to use if previous methods have index. failed. 12 A demonstration of the problel»- is shown in (37). 13 Cf. R.FUGMANN (12), W.GODERT (37), and A.PITER· The retrieval language must, in addition to Boolean NICK (34) operators, also pennit the usc of syntactic aids as well as 14 'Suggestive prompts' arc messages from the system, show­ numeric area scanning. The retrieval process takes place ingthe user which are the possibilities at this point of his dia­ logue. They are indirect help measures, not necessarily to be as a rule via the display of the indexed items. In that way followed. cr. (45).

74 Int. Classif. 17 (1990) No.2 · Gndert/Horny: Subject Access Elements in OPACs 15 Masking and truncation are mostly used synonymously. In (22) Oberhauscr, 0.: Klassifikationin Online-Informationssys­ this connection, however, we would like to distinguish as fol� temen.lnLClassif. 13(1986)p.79-87 lows: Masking has the advantage that the display ends with (23) Svenonius, E.: Usc of classification in online retrieval. the word which is the last onc to show the word stem put in. Libr.Resources & Techn.Serv. 27(1983)p. 76-80 In truncation fu rther words are displayed and it is possible (24) Hill, J.: Online classification number access - some practi­ to go on with browsing further to the word stem put in. cal considerations. 1.Aead.Librnship 10(1984)p.17-22 16 In Walt CRAWFORD's book (55) numerous possibilities (25) Huestis, J.: Clustering LC classification numbers in an on­ of variation arc given. line catalog for improved browsability. Inform.TechnoL& 17 See the detailed explanations in (56) as well as HILDRETH Librs. 7(1988)p.381-393 in (45) (26) Williamson, N.: The Library of Congress Classification 18 About this A.LIPOW wrote very aptly: "Since making the and the computer research in progress. Int.Catalog.& Bib­ system user-friendly means that the patron is always liogr.ControI 18(1989)p.8-12 presented with a manageable set of choices, and since you (27) Funk, H., Loth, K.: Sachabfrage im ETHICS aufder Basis can not write a set ofchoices for every conceivable situation, der UDK. In: Wissensorg.im Wandcl: Dezimalklassifika­ "user-friendly" to me is synonymous with limited services", tion, Thesaurusfragcn, Warenklassifikation. Proc.Aachen see (56). 1987. Frankfurt: Indeks Ver1.1988. p.103-1 14 (28) Deutsches Institut fUr Normung. NABD: DIN 31 623 In­ dexierung zur inhaltlichen ErschlieBung von Dokumenten. Teil 2: GleichordnendeIndexierungmit Deskriptoren. Ber­ References lin: Beuth Verl.1983. 17p. (29) Korner, H.: Syntax und Gewichtung in Informations­ (1) Besant, L.: Users of public access catalogs want sophisti­ sprachen. Nachr.Dok. 36(1985)p.82-1 00 cated subject access. Amer.Libr.13(1982)p. 160 (30) Fugmann, R.: Die Aufgabenstellung zwischen Wortschatz (2) Kaske, N.K., Sanders, N.P.: On-line subject access: the und Grammatik in einer Indexsprache. In: Datenbasen, human side of the problem. RQ Fall (1980)p.52-58 Datenbanken, Netzwerke. Munchen 1979. Bd.I, p.,67-93 (3) Markey, K.: Thus spake the OPAC user. Inform.Tech­ (31) Fayen, E.: The online catalog: improving access to library no!.& Librs. 2(1983)p.381-387 materials. White Plains 1983. p.76 (4) Markey, K.: Users and the online catalog: subject access (32) Austin, D.: Automatisierung in del' SachersehlieBung dcr problems. In: Matthews, J.R(Ed.): The impact of online British Library. Bibliothek 8(1984)p.45-57 catalogs. London 1986. p.35-69 (33) Dubois, C.: Free text vs controlled vocabulary. Online (5) Moore, C.W.: User reactions to online catalogs: anexplora­ Rev. II(l987)p.243-253 tory study. College & Res.Librs. 42(1981)p.295-302 (34) Piternick, A.: Searching vocabularies a developing ca­ (6) Norden, D., Lawrence, O.H.: Public terminal use in an on­ tegory of online search tools. Online Rev. 8(1984)p.441-449 line catalog: some preliminary rcsults. College & (35) Binder, W.: Quo vadis Onlinc�Katalog. ABI.­ Res. Librs. 42(1981)p. 308-316 Techn.9(1989)p.I-20 (7) Specht, J.: Patron use of an online-circulation system in (36) Jones, C: Subject access: Report ofa meeting sponsored by known-item searching. J.Amer.Soc.Inform.Sci. 31 (1980) the Council on Library Resources, Dublin, OH, June 7-9, p. 335-346 1982. (K.Russel, Comp.& Ed.). Washington, DC 1982. (8) Steinberg, D., Metz, P.: User response to and knowledge p.6-8 about an online catalog. College & Res.Librs. 45 (1984) p. (37) Godert, W.: Online Katalog und bibliothekarische Inhalts­ 45-66 erschlieBung. In: Haase, Y.(Ed.): Reden und Vortdige. (9) Matthews, J.R.(Ed.): Using online catalogs: a nationwide 77.DLBibliothekartag, Augsburg 1987. Frankfurt 1988. survey; a report of a study sponsored by the Council on Li­ ZfBB Sonderheft 46, p.279-302 brary Resources. New York 1983. 255p. ISBN (38) Rcynolds, D.: Library automation: issues and applica­ 0-918212-76-0 tions. New York 1985. p.438 (10) Markey,K.: Subject searching in library catalogs: before (39) Laisiepen, K. et al: Grundlagen der praktisehen Informa­ and after the introduction of online catalogs. Dublin: tion und Dokumentation: eine Einfiihrung. 2.vollig neu­ OCLC 1984. p.77 bearb.AuO. Munchen: K.G.SaurVer1.1980. p.532 (11) Tagliacozzo, R., Kochen, M.: Information seeking beha­ (40) Hildreth, CR.: To Boolean or not to Boolean. In­ vior of catalog users. Inform.Storage & Retrieval fo rm.Teehno!.& Librs. 2(1983)p.235-237 6(1970)p.363-381 (41) Gaus, W.: Dokumentations- und Ordnungslehre: (12) Fugmann, R.: Theoretische Grundlagen del' Indexierung­ Lehrbuch fUr die Theoric und Praxis des Information Re­ spraxis. In: Anwend. in der Klassifikation. Proc.8.Jahresta­ trieval. Berlin 1983. p.265 gung der Ges.f.Klassif. Frankfurt: Indeks Ver1.1985. Stu­ (42) Gebhardt, F.: Dokumentationssysteme. Berlin 1981. dien z.Klassif.Bd.14, p.42-56 p.116-117 (13) Hildreth, CR.: Pursuing the ideal: generations of online (43) Salton, G., McGill, M.: Information retrieval: Grund­ catalogs. In: Avency, B. (Ed.): Online catalogs, online ref­ legendes fUr Informationswissenschaftler. Hamburg 1987. erences: converging trends. Proc.Libr.& Inform.Tech­ p.31-33 no1.Assoc.Preconf.Institute, June 23-24, 1983, Los (44) Lawrence, G.: System features for subject access in the on­ Angeles. Chicago: ALA 1984. p.31-56 line calalog. Libr.Resources & Teehn.Serv. 29(1985) (14) Mandel, CA.: Enriching the library catalog record for sub­ p. I 6-23 ject access. Libr.Resources & Techn.Serv.29(1985)p.5-15 (45) Hildreth, CR.: Online public access calalogs: the user inter­ (15) GDdert, W.: Syntax von Dokumentationssprachen im On­ face. Dublin, OH 1982. p.65 line-Katalog. In: Wissensorg.im Wandel: Dezimalklassifi­ (46) Williamson, N.: Subject access in the online environment. kation, Thesaurusfragen, Warenklassifikation. Proc. II. In: Advances in Librarianship 13(1984)p.63 Jahrestag. d. Ges. f. Klassif., Aachen 1987. Frankfurt: 111- (47) Crawford, W.: Patron access: issues for online catalogs. deks Ver1.1988. � Studien z.Klassif.Bd.18, p.103-1 14 Boston 1987. p.168-169 (16) Cochrane, P.: Classificationas a user's tool in online public (48) Kinsella, J.: Prospects for browsing: experimental ap­ access catalogs. In: Cochrane, P.: Redesign ofcatalogs and proaches to the presentation of brief entries and the design indexes for improved online subject access. Phoenix, AZ of "browse screens". In: Helal, A. (Ed.): Future of online 1985. p.382-390 catalogues: Essen Symp. 30 Sept.-3 Oct.1985. Essen 1986. (17) Godert, W.: Bibliothekarische Klassifikationstheorie und p.227-241 on-line Kataloge. Bibliothck 11(1987)p. 152-166 (49) Congreve, J.: Problems of subject access. Program (18) Godert, W.: Klassifikationssystems und Online-Katalog. 20(1986)p.204-210 Z. Bib!.wes. u. Bibliogr.34( 1987)p.185-195 (50) Corey, J.: Search retrieval options. In: Aveney, B.: Online (19) Hermes, H.l.: OPAC, Verbund, DK. Biblos 36 (1987) p. catalog design issues: a series of discussions. Washington, 162-178 DC 1984. p.23-66 (20) Markey, K.: Subject searching experiences and needs ofon­ (51) Massicotte, M.: Improved browsable displays for online line catalogs users: implication for library classification. subject access. Inform.Technol.& Librs. 7(1988)p.373-380 Libr. Resources & Techn.Serv .29( 1 985)pO.34-51 (52) Janosky, B., Smith, P., Hildreth, C: Online library catalog (21) Markey, K.: Class number searching in an experimental on� systems: an anlysis of user errors. Int.J.Man-machine line catalog. l nLClassif. 13(1986)p.142-150 Studies 25(1986)p.573-592 lnt. Classif. 17 (1990) No.2 -GodertfHorny: Subject Access Elements in OPACs 75 (53) Markey, K.: Alphabetical searching in an online-catalog. (65) Mandel, C, I-Icrschman, J.: Online subject access: enhanc­ J.Acad.Librnship 14(1989)p.353-360 ing the library catalog. J .Acad.Librnship 9(1983)p. 148� 155 (54) Bates, M.: Subject access in online catalogs: a design (66) Eversberg, B.: Bemerkungen zum Sachzugriff im Online­ model. J.ASIS 37(l986)p.357-376 Publikumskatalog. In: Mitt.bI.d.Bibl.in Niedersachsen (55) Crawford, W.: Bibliographic displays in the online catalog. (1987)No.66, p.5-9 White Plains 1986. 359p. (67) Schllelling, H.: Katalogvcrbund, Fremddatennutzung und (56) Budnitz, H.: Die Gestaltung der Benutzeroberflache von Online-Benutzerkatalog: Aspekte einer Neuorientierung Online-Benutzerkatalogen. ABI-Techn.8( 1988)p.307 -326 der SaeherschlieBung. Libri 38(1988)p.237-256 (56a)Lipow, A.: Practical considerations of the current capa­ (68) Schllelling, n.: M6g1ichkeiten nnd Grenzen der Sacher­ bilities of subject access in online catalogs. Libr.Resources schlieBung im Hinblick auf Olllille-Benutzerkatalogc: An­ & Techn.serv. 27(1983)p.81-87 merkungen zur Situation in del" Bundesrepublik Deutsch­ (57) Arret, L.: Can online catalogs be too easy? Amer. land. Liber Bull.32/33(1989)p.l42-162 Librs. 16(1986)p. I 18-120 (69) Wiegand, G.: Anmerkungen zuden Empfehlungen des Wis� (58) Hildreth, C.: Beyond Boolean: designing the next gener­ senschaftsrates zur retrospektiven Katalogisierung an wis­ ation of online catalogs. Libr.Trends 35(I987)p.647-667 scnschaftlichcn Bibliothekell. Mitl. bI. Verb. Bib!. Land. (59) Lawrence, G.: System fe atures for subject access in thc on­ NRW, NF 39 (1989) No.1, p.12-15 line catalog. Libr.Resources & Techn.Serv .29(1985)p.1 6-23 (70) Bates, M.: Designing online catalog subject access to meet (60) Walker, S., Jones, R.: Improving subject retrieval in online user needs. Paper presented at the IFLA Congress 1989, catalogs. YoU: Stemming, automatic spelling correction Paris. (37-CLASS-2-E) and cross-reference tables. London 1987. I50p. (71) Cochrane, P.A.: Subject access - free or controlled? The (6l) Bookstein, A.: Probability and fuzzy-set applications to in­ Case of Papua New Guinea. Tn: Cochrane, P.A.: Redesign fo rmation retrievaL Ann.Rev.lnform.sei.& TechnoL of catalogs and indexes for improved online subject access. 20(1985)po. 117-151 Phoenix, AZ 1985. p.251-266 (62) Willett, P.: Ranked output searching in textual and structu­ (72) Akeroyd, J.: Information seeking in online catalogues. ral databases. In: 9th Int.Oniine Meeting: London, 3-5 J.Dae.46(1990)p.33-52 Dec.1985. Oxford, GB: Learned Inform.(Europe) 1985. (73) Aitchison, J., Gilchrist, A.: Thesaurus construction: a prac­ p.343-353 tical manual. London: Aslib 1987. (63) Mitrev, N., Venner, G., Walker, S.: Designing an online public access catalogue: OKAPI, a catalogue on a local Ms.Silke Horny, Universitat Konstanz, Siidwestdeutscher Bib­ area network. London 1985. 254p. liotheksverbulld, Postfach 55 60, D-7750 Konstanz. (64) Settel, B., Cochrane, P.: Augmenting subject descriptions Prof.Winfried G6dert, Fachhochschule fUr Bibliotheks- und for books in online catalogs. In: Cochrane, P.: Redesign of Dokumentationswesen, Claudiusstr. I, D-5000 Knin I catalogs and indexes for improved subject access. Phoenix, AZ 1985. p.461-475

Meeting on Concept Relationships As in previous years, Prof.Rudolf WILLE and his re­ schaflen. - R.HULE, Innsbruck: Pragmatische Begriff­ search group of the Mathematical Inslitute, Technical sanalyse I: Theoretische Grundlagen. - *I.ZELGER, University of Darmstadt invited on behalf of his Re­ Innsbruck: Pragmatische Begriffsanalyse II: Vorliiufige search Group and the Special Interest Group on Concept Erfahrungen und Ergebnisse. - *B.GANTER, Darm­ Analysis of the German Society for Classification to par­ stadt: Begriffssysteme mit Symmetrie. - *F.VOGT, ticipate in an introductory courSe on formal concept ana­ Darmstadt: Datenanalyse auf der Grundlage einer be­ lysis (March 8-9) and afterwards (March 9-11) in a meet­ grifllichen Datei. - *R.FRITZSCHE, Halle: Merkmal- ing on concept relationships. According to the list of par­ simplikationen bei der Erdfernerkundung. ticipants, some 74 persons attended the latter at which 18 *R.WILLE, Darmstadt: Begriflliche Wissenssysteme. papers were presented and discussed. However, there P.LUKSCH, Darmstadt: Das automatische Lernsystem were only a few concerned expressedly with concept rela­ PINOCCHIO aus begriffsanalytischer Sichl. W.L.FI­ tionships in a more or less philosophical-logical, psycho­ SCHER, Erlangen: Topologische Invarianten von Inzi­ logical and linguistic way; thus there scemed not to be denzmatrizen und Anwendungen. - K.E.WOLFF, something really new in the sense which is needed in the Darmstadt: Anwendungen subjektiver MaBe del' begriff­ construction of conceptual or classification systems as lichen Skalierung. - N.SPANGENBERG, Frankfurt: such. There were, however, a few papers with strikingly Begriflliche Erhebungstechniken in der Familienthera­ new ideas which ought to be considered in the develop­ pie. - B.RUTTINGER, A.SOURISSEAUX, Darm­ ment of the field of concept analysis. It is to be hoped that stadt: Komponentcn der Arbeitsqualitiit. *G.SCHAE­ the organizers ofthis meeting would do something about FER, Hamburg: Der Begriff 'Leben' und seine Relation their publication. zu Nachbarbegriffcn. - J.sCHAFER, Darmstadt: Ethik In the fo llowing list of papers (in the sequence of their und Pathologie del' Metapher. - *K.MUDERSBACH, presentation) I will mark those with an asterisk of which I Heidelberg: Fehlentwicklungen in del' Extension-Inten­ thought they contain important research findings and sion-Unterscheidung. - G.RAHMSTORF, Heidelberg: should be published soon: I. DAHLBERG, Frankfurt: Vergleich von Begriffen und Relationcn aus linguisti­ Kants analytische und synthetischc U rteile und ihre be­ scher Perspektive. -* J .HELLER, Regensburg: Zur meB­ griftlichcn Relationen. - E.TEGTMEIER, Mannheim: theoretischen Begriindung del' Rcprasentation von Be­ Verwandtschaft und Unvertraglichkeit zwischen Eigen- griffsrelationen in del' Psycholinguistik. I. Dahlberg

76 Int. Classif. 17 (1990) No.2 Gndert/Horny: Subject Access Elements in OPACs Jaana Kristensen Kalervo Jarvelin University of Tampere, Finland

The Effectiveness of a Search­ ing Thesaurus in Free-Text Searchi ng in a Full-Text Database

Kristensen, J., Jarveiin, K.: The effectiveness of a searching the­ operator) and the applicability offree text searching. We saurus in frce-text searching in a full-text database. shall use the term controlled vocabulary searching when InLClassif.17(1990)No.2, p.77-84, 30 refs. the search terms are picked from a controlled vocabu­ Synonyms, antonyms, quasi-synonyms and homonyms are typi­ lary. The term free text searching is used when the search cal fe atures of natural language which cause difficulticsin free­ terms are picked from any available source (e.g. the text searching in fu ll-text databases. Searching thesauri which supply synonyms, antonyms and quasi-synonyms for given user's mind). The term bibliographic retrieval denotes re­ natural language terms have been suggested as a search aid to trieval based on the bibliographic data fields; either con­ overcome these problems. To test these suggestions, a small trolled vocabulary searching or free text searching, or searching thesaurus was constructed for Finnish newspaper ar­ ticles on economic issues and a set of test searches were con­ both may be used. (Obviously, in most of the cases, con­ ducted and analyzed in a large Finnish newspaper article data­ trolled vocabulary searching is applied on the index term base. The original queries were elicited from journalists who field.)The term full-text retrieval denotes retrieval based also evaluated the rclevance of the search results. Each query was searched in three distinct modes: basic search, synonym on the document texts; in principle, either controlled vo­ search and related term search. The basic searches contained cabulary searching or free text searching, or both may be only terms included in the journalists' original query statement. used (although the former is not quite relevant). Both bib­ The logic of the basic search was developed by the researchers. In the synonym searches, the terms of the basic search were ex­ liographic retrieval and full-text retrieval are possible in tended by disjunctions of the synonyms given by the searching TDBs provided that the necessary data fields are avail­ thesaurus without modifying the ovcrall logic of the basic search. In the related term searches, the disjunctions in the sy­ able. nonym searches were further extended by disjunctions of related The relative effectiveness and associated potentials tcrms (quasi synonyms) given by the searching thesaurus again and problems (e.g. indexing costs) of controlled vocabu­ without modifying the overall logic of the synonym search. The searches in each mode were analyzed W.r.t. relative recall and lary searching and free text searching have been dis­ precision by setting the relative recall ofthe third mode searches cussed in the literature (e.g. [IJ [2J [5J [6J [7J [8J [18J [19J [26J to 100 %. The average relative recall in the first modesearches [27J [30]). There does not seem to be one single best was ca. 45 % and in the second ca. 82 %. The average precision values were for the modes one, two and three were 51 %,41 % method -- controlled vocabulary searching and free text and 33 %, respectively. The substitution of synonyms shows a searching rather augment each other [6] [30J. The considerable increase in recall with a negligible decline in preci­ strengths and weaknesses of each method are well­ sion. (Author) known (e.g. [25]). Relatively recent empirical investiga­ tions on the performance of full-text retrieval in large operational TDBs are the evaluations by TENOPIR [30J and BLAIR and MARON [IJ [2J. The general argument by Tenopir has been that fu ll-text retrieval (through free t.Introduetion text searching) improves recall at the cost of deteriora­ The evolution of information technology has made ting precision. Blair and Maron argue that even the re­ the electronic storage of very large text collections call tends to be low and that automatic indexing methods possible. Such collcctions, text databases (or full-text da­ are not very promising. The diffe rences in the findings tabases), often contain in a computer readable form the can be explained at least in part by differing approaches, whole text, or at least large excerpts, of documents col­ different types of texts and evaluation criteria [21 J [28J. lected for some purpose. Text databases (briefly TDBs) SALTON challenges the results against automatic index­ may serve the daily routines of an organization, distribu­ ing [22]. Tools aiding in the identification of relevant tion of business information , or the communication of search terms for free text searching have not been evalu­ scientific information. TDBs are more relevant to the ated in these studies. end-users than the conventional bibliographic databases Free text searching requires that sufficientand appro­ because the fo rmer contain source data rather than surro­ priate synonyms, quasi-synonyms and antonyms afC gates. [4J [12J [13J [29J identified for the concepts describing the search topic. Information retrieval in publicly available TDBs is LANCASTER pointed out that any means, incl. existing based to a greatdegree on the same methods as the retrie­ controlled vocabularies, can be used as aids for term val from bibliographic databases. Differences emerge in identification [15J. Later the use of searching vocabu­ the importance of some operators (e.g. the proximity laries or searching thesauri have been suggested for this

Int. Classif. l7 (1990) No.2 -- KristensenJJiirvelin: Thesaurus in rree-text searching 77 purpose [16] [20]. A searching thesaurus is not used in the guage [II] [14]. indexing of the database contents. It is used only in the re­ Although Finnish is a highly inflectional language, trieval phase. In contrast to conventional thesauri, it our study is not language bound, in principle. Sy­ docs not seek to standardize the term choices by the sear­ nonymy, quasi-synonymy and other alternative forms of cher but rather provides alternatives (synonyms, quasi­ expression are fe atures of all natural languages, not just synonyms, antonyms and related terms) to the terms the Finnish. Only ifit can be shown, that there are significant searcher has in his/her mind. Thus a searching thesaurus differences in the number of synonyms etc. and in the fre­ aims at providing some ofthe benefits of controlled voca­ quency of their use between different languages, our re­ bularies without the cost of indexing. sults are language bound to some minor degree. If so, our In this study, we evaluate the effectiveness ora search­ results are still indicative on the effects and value of ing thesaurus in terms orits effects on thcrccall and preci­ searching thesauri. sion of search results. In particular, we want to find out its usability as an end-user tool in information retrieval. 2.2. The test searching thesaurus We do not know of earlier empirical, statistical tests on A newspaper text is different from scientificand techni­ the effectiveness of searching thesauri. Our approach can cal text both in vocabulary and in structure. There was no be characterized in terms of sources of the test environ­ suitable documentation language available to serve as ment, the test thesaurus, the test searches and search re­ the basis for a searching thesaurus. The vocabulary and sult evaluation: structure of Finnish subject heading lists and general (a) The test environment is a TDB containing some classifications(e.g. the UDq were unsuitable. Therefore 34,000 Finnish newspaper articles operated under a small test thesaurus was constructed for the study on the BASIS information retrieval and data manage­ the basis ofterms occurring in newspaper articles on econ­ ment system. omic issues. Thus the test thesaurus has a literary war­ (b) A small test thesaurus was constructed for the study rant. As a fu rther advantage, it has also a user warrant be­ on the basis of terms occurring in newspaper articles cause journalists were the sole sources of both the articles on economic issues. Thus the test thesaurus is literary in the database and the test searches. warrant. The test searching thesaurus was organized by follow­ (c) Original quelies (on economic issues) were elicited ing the guidelines for thesaurus construction [10]. The fr om journalists. Each query was searched in three structure of the thesaurus is quite conventional. How­ distinct modes: basic search, synonym search and re­ ever, there is no distinction between descriptors and lated term search. The basic searches contained only entry terms: all terms belonging to the same equivalence terms included in the journalists' original query state­ class are treated equally becausc they were all equally ment. The logic of the basic scarch was developed by possible search terms. The relationships between the the researcher. In the synonym (and related term) terms were the usual ones: broader term, narrower term, searches, the terms of the basic search were extended synonym, related term. The thesaurus contained 180 dif­ by (inclusive) disjunctions of the synonyms (and the ferent concepts represented by altogether 328 different related terms) given by the searching thesaurus with­ terms. Some test thesaurus excerpts are given in the Ap­ out modifying the ovcrall logic of the basic search. pendix, (d) Each journalist also evaluated the relevance of each article retrieved by the searches for his/her query 2.3. The test searches statement in three classes (non-relevant, maybe rele­ The original queries (on economic issues) were elicited vant, relevant). from journalists working on economic issues. The re­ The chapter organization of this paper is as follows. search setting was explained to them and they were asked Materials and methods are described in detail in Chapter to state queries they would execute in a text retrieval 2; the findings are reported in Chapter 3; a discussion is envi­ ronment. Five journalists participated in the test and al­ given in Chapter 4; and the conclusions fo llow in Chap­ together 30 realistic queries were collected from them. ter 5. The journalists' own words were the most compact ex­ pressions of their needs in free-text terms. Each separate 2. Methods and Materials concept a journalist had in his (her) mind was expressed 2.1. The test environment by the terms the journalist used. This was used as the The test environment is a TDB containing Finnish basis for the basic mode searches which contained only newspaper articles operated under the BASIS [17] infor­ terms included in the journalists' original query state­ mation retrieval and data management system. The data­ ment. base contains the articles published in the newspaper Sat­ The logic of the basic search was developed by the re­ akunnan Kansa (published in Pori, Finland) from May searcher. The terms representing different concepts were 31, 1987 to September 7, 1987, altogether some 34,000 ar­ connected by the conjunction operator Cand'). When a ticles. The database is large enough for providing realis­ concept was represented by a phrase, the words in the tic findings(cf. [1 D. In thc basic index ofthe database, the phrase were connected by a proximity operator restrict­ terms were given in their basic form only irrespective of ing the word occurrences to a single paragraph. For their inflected occurrences in the text. This was achieved example, if a journalist asks for articles on plans for nu­ by preprocessing the basic index entries by the morpho­ clear energy in the paper industry, the concepts were nu­ logical analysis program MORFO for the Finnish lan- clear energy and paper industry, and the basic search

78 lnt. Classif. 17 (1990) No.2 - Kristenscn/liirvclin: Thesaurus in rrec-text scarching could be nuclear energy AND paper industry. (NB. the or no related terms provided by the test searching the­ proximity operator for the phrase 'nuclear energy' is not saurus. The corresponding search mode was skipped in given explicitly). The set of articles retrieved by this for­ such cases. The synonym extension was possible for 23 mulation is denoted below by B. original queries and the related term extension for 21 The problems due to synomyms, quasi-synonyms and queries. There were 18 queries for which it was possible other flexibility of natural language in full text retrieval to form searches in all three search modes. are well known. They tend to appear also in indexing ex­ In the data analysis, also the net effect of query exten­ periments. For example, IIVONEN has investigated sion types recognized. The net effect of a synonym search inter-indexer consistency in free-text and controlled vo­ is the set of articles retrieved by the synonym search and cabulary indexing. She found out that different indexers, not retrieved by the corresponding basic search. In the while being inconsistent in their term choices, tend to above example, the net effect of the synonym search is S­ choose terms related by equivalence or associative rela­ B. The net effect of a related term search is the set of ar­ tions [9]. The same is probably true of journalists writing ticles retrieved by the related term search and not re­ newspaper articles: while writing on the same topic, dif­ trieved by the corresponding synonym search. In the ferent journalists use different terms which are likely to above example, the net effect of the related term search is be related by equivalence or associative relations. To im­ R - S. By differentiating the net results, it was possible to prove recall in full text searching, alternative forms of ex­ analyze the quality of the search extensions in detail. pression for each concept must be accounted fOf. We stu­ died two extension types of the basic searches: synonym 2.4. Search result evaluation extensions and related term extensions. Thiswas the ra­ The researcher removed obviously irrelevent articles tionale for the two additional search modes for each from the results of the related term searches. Thereafter query: the synonym search and the related term search, eachjournalist also evaluated the relevance of the remain­ which were also developed by the researcher following ing articles retrieved by the related term search for his/ the guidance by the searching thesaurus. her query statement. (NB: The articles removed by the re­ In the synonym searches, the terms of the basic search searcher are counted as irrelevant in the statistics). The re­ were extended by (inclusive) disjunctions of the sy­ levance evaluation consisted of three classes: irrelevant, nonyms given by the searching thesaurus without modif­ maybe relevant, relevant. Only articles in the last class ying the overall logic of the basic search. Continuing the were counted as relevant in the recall and precision calcu­ example above, the synonym search could be lations. One or more relevant articles were retrieved for (nuclear energy OR nuclear power OR atomic energy OR 26 different original queries, i.e. there were four queries atomic power) AND (paper industry OR papermills) for which none of the search modes retrieved any rele­ assuming that the synonyms were provided by the vant articles. These four original queries were excluded searching thesaurus. This search is still an intersection of from the analysis. Altogether 70 searches were con­ two concepts, i.e. the overall logic remains the same as in ducted for the 26 original queries. The 70 searches the basic search, but now each concept is represented by yielded altogether 248 relevant and 493 irrelevant articles. an inclusive disjunction for terms. The set of articles re­ The search results were evaluated for recall and preci­ trieved by this fo rmulation is denoted below by S. Ob­ sion. The difficultyof estimating the recall of searches in viously, the result of a synonym search contains the re­ operational databases of realistic size is well known. In sult ofthe corresponding basic search, i.e. S ;2 B. our setting it was not necessary to estimate the absolute In the related term searches, the disjunctions in the sy­ recall of the searches. Because we were interested in the nonym searches were further extended by (inclusive) dis­ comparison of the different search modes rather than junctions of related terms (quasi synonyms) given by the their absolute performance and because the results from searching thesaurus again without modifying the overall each related term search contained the results from the logic of the synonym search. Continuing the example other search modes for the same original query, it was above, the related term search could be possible to compute relative recall figures for these sear­ (nuclear energy OR nuclear power OR atomic energy OR ches. The relative recall of the related term search was set atomic power OR nuclear power plants OR nuclear power sta­ tions OR nuclear energy production OR atomic energy produc­ at 100 % for each original query. When the related term tion) AND (paper industry OR paper mills OR paper OR cellu­ extension was not possible, the relative recall of the sy­ lose OR carton OR cardboard) nonym search was set at 100 %. again assuming that the related terms were were pro­ The recall figuresof the other search modes were then vided by the searching thesaurus. The set of articles re­ counted on the basis of their result sizes. For example, trieved by this formulation is denoted below by R. Ob­ continuing the sample query on nuclear energy and viously, the result of a related term search contains the re­ paper industry, recall the search result sets B, S, and R. sult of the corresponding synonym search, i.e. R 2 S. Let R + be the set of relevant articles in R. Then the sets Thus the results of the synonym searches contained of relevant articles in the basic and synonym searches are the results of the basic searches; the results of the related B n R + and S n R + , respectively and the number of term searches contained the results of the synonym sear­ relevant articles in each search 1 B (j R + I, 1 S nR + I, ches, i.e. R;2S ;2 B. The full results of the related term and 1 R + I. Thus the relative recall ofthe basic search is searches were printed. The access numbers of the articles (I B n R+ 1/1 R+ I) * 100 % and the relative recall of the in the results of the basic and synomyn searches were re­ synonym search is ( 1 S n R+ 1/1 R+ I) * 100 %. We ob- corded. For some queries there were either no synonyms . tained therefore no data on how many articles were lnt. Classif. 17 (1990) No.2·- Kristensen/IiirveIin: Thesaurus in free-text searching 79 missed by all the searches but we got data on the relative Table �. Average percentages of relative recall and precision for the three search modes : the baSIC search, the synonym search and the related term search (all 26 original queries). performance of the search modes. In other words, we are able to compare the effectiveness afthe test searching the­ Avcrage % Basic Search Synonym Search Relatcd Term Search saurus as the source of suitable synonyms and related (N"'26) (N=o23) (N"'21) terms. The false drops were also analyzed in order to findout Relative 45,2 81,8 100,0 Recall their types and causes. We classified the false drops into Precision 50,6 40,6 33,1 those caused by a) article text, b) search strategy, and c) the test thesaurus. This classification is conventional Table 2. Average percentages of relative recall and precision in the subset of 18 queries (e.g. [15] [16]). The class for 'indexing errors' is non-rele­ for which allthe three search modes : the basic search, the synonym search and the related vant in our setting. term search were evaluated. False drops caused by article text may be due to: Average % al) search terms not being related in the text, Basic Search Synonym Search Relatcd Term Search a2)incorrcct relationship between the search terms, a3)homonyms, Relative 44,4 76,8 100,0 Recall a4) the tcxt dealing with the search topic only marginally 35,0 or in a wrong way. Precision 48,5 40,2 False drops caused by search strategy may be due to : bl)failing to account for some concept in the original query, b2)representing some concept by a too broad term, b3)rcprcscnting some concept by an incorrect term. ticles retrieved by average number of articles retrieved Conceptual correctness of the test thesaurus affects the and mUltiplying this by 100; e.g. in the basic mode aver­ number of false drops. Incorrect or spurious term rela­ age precision � 50,6 # 100 x 4,0 / 9,3 � 43,0). tionships in the thesaurus cause false drops. The analysis of the subset of those 18 original queries for which all three search modes (see Table 2) 3. Findings shows similar results. The increase in the relative recall seems here to be slightly better than in the set of 26 orig­ 3.1. Overall recall and precision effects inal queries. Precision seems to deteriorate slightly fa ster Th,e·test searching thesaurus was relatively small in the in this subset than among all original queries. The aver­ conceptual coverage. Therefore the thesaurus could not ages of total number of articles retrieved and relevant ar­ give terms for extending the basic mode search either by ticles retrieved werc 9,2 and 4,1 for the basic search synonyms or by related terms for some original queries. mode, 18,8 and 7,6 for the synonym search and 28,4 and Recall and precision are therefore evaluated separately 9,1 for the related term search. for all the 26 original queries (with 26 basic searches, 23 synonym searches and 21 related term searches) and for 3.2 The net effect of query extension types the subset of those 18 original queries for which all thrce search modes were available. The net effect of query extension types is shown in The results for all 26 original queries are given in Table Table 3. The net effect of the synonym scarches was such 1. These findings indicate that the precision of the basic that they brought almost one third ofthe total relevant ar­ search was about 50% and it deteriorated by 10% in the ticles and almost every third article retrieved was judged synonym searches and by some 17% in the related term relevant. The net effect of the related term searches was searches. As expected, the relative recall figures im­ such that they brought almost one quarter of the total proved when the precision figures deteriorated. The relevant articles but only less than one out of six articles basic mode searches retrieved, on the average, only 45% retrieved was judged relevant. The averages of total num­ of the articles considered relevant in the related terms ber of articles retrieved and relevant articles retrieved searches. The inclusion of synonyms improved recall by were 9,2and4,1 for the basic search mode, 9,5 and 3,5 for 36% (at the cost of 10% decline in precision) and the in­ the synonym extension and 9,7 and 1,5 for the related clusion of related terms by additional 18% (at the cost of term extension. fu rther 17% decline in precision). These effects in Although each extension type retrieved roughly relative recall and precision are statistically significant equally many new articles, the number of relevant ar­ both between the basic search and the synonym search ticles retrieved deteriorated considerably from the basic and between the basic search and the related term search search to the relatcd term extension. This is clear also (see Table 4 in Section 3.3 below for significancetests) . from the precision figures. (NB: Here, also, the precision The total number of articles retrieved for all 26 orig­ figuresare averages of precision figuresof individual sear­ inal queries varied fr om 1 to 51 per search. The averages ches while the counts are sums of total (relevant) articles of total number of articles retrieved and relevant articles divided by the number of searches (18). Therefore preci­ retrieved were 9,3 and 4,0 for the basic search mode 21 3 sion # 100 x relevant articles / total articles). and 8,9 fo r the synonym search and 28,0 and 8,6 [or ther�­ lated term search. (NB: the average of the precision 3.3 Statistical significance of the findings figures of individual searches is different from the figure Statistical significance of the findings are reported in obtained by dividing the average number of relevant ar- Table 4. Significanceis tested both with the t-test and the

80 Tnt. Ci

the significance of the relative recall and precision dif­ Basic Search vs. 4.963"� 3,530"· ferences for these 23 searches. Synonym Search (Neo23) Table 4 indicates that the improvements in relative re­ Basic Search vs. 9,459"� 4,218"· Related Term Search (N",21) call from the basic search to synonym search and related Basic Search vs. 1,049 .,.a) term search are statistically significant with a very small Synonym Extension (NeoI8)

risk level (p < 0,001). Also the deterioration of the preci­ Basic Search vs. 1,665 1,787* sion figuresfrom the basic search to synonym search and Related Term Extension (N"d8) related term search are statistically significant (p < 0,0 1). PRECISION With this data, the deterioration of the relative recall in Basic Search vs. 2,771*' 2,875·· the net results of query extension types arc not statisti­ Synonym Search (N"'23) cally significant. According to the Wilcoxon test, how­ Basic Search vs. 2.676'· ever, the deterioration ofthe relative recall fr om the basic Related Term Search (Noe2l)

search to the related term extension is almost significant Basic Search vs, 2.274* 2.275* Synonym Extension (Neo18) (p < 0,05). In contrast, even this small data set shows a statistically very significant (p <0,001) deterioration of Basic Search vs. 4,177"· 3,087"· Related Term Extension (Neo18) precision from the basic search to the related term exten­

sion. The difference in precision between the basic search "' p

Table 3. Average counts or total articles and relevant articles retrieved and average percentages of proportional recall and predsion in the net results the basic search and the search extensions : the synonym extension and the related term extension. (N eo 18)

Averages

Counts Percentages Extension Type Total art. Relevant Arl. Relative Recall Precision cide. For example, the expressions "ei lainkaan" [not at all] and "ei milleen lailla" [in no way] match two inflected Basic Search 9,2 4,1 44,4 48,5 form instances of the word "laki" [law] . If one is search­ Synonym 9,5 3,5 32,4 30.7 Extension ing for law issues, one has to account for inflection either by searching truncated terms "laki- or lai- or laei-" to Related Term 9.7 1,5 23,2 12,9 Extension cover inflection or by recognizing the basic forms of the text words for a basic form basic index and searching then for the basic form "laki". The disjunct "lai-" in the fo rmer case matches the irrelevant strings. In the latter 3.4. Analysis of false drops case, the basic form algorithm recognizes the noun "laki" The 493 irrelevant articles found by all the searches as one possible basic form for the strings "lain- kaan" were analyzed as to the reasons of irrelevance. The results and "lailla". By searching only for "laki-" one would cer­ of this analysis are tabulated in Table 5. The most com­ tainly miss essential articles on the topic. The only mon sources for irrelevance were search terms not being related in the article and search topic being discussed only marginally in the article. These accounted for 80% of false drops. The use of proximity operators would have reduced the former type offalse drops effectively. In

this study, word proximity operators were used only for Table S. The number of original queries and the number of different searches affected precoordinated phrases of the test thesaurus. The hy false drops and the number of false drops for differe nt sources of false drops. Boolean intersection was used between terms reprcscnt­

ing different concepts. The latter type of fa lse drops Source of false drop No. of queries No. of searches No. of articles could be reduced by setting term frequency thrcsholds for the search terms: articles with search term frequencies Terms not related 21 40 221 below the threshold would be rejected. In this study, term Incorrect term relation· frequency thresholds were not tested. ships in document 16 22 52 Homonymy did not cause many false drops. However, Homonymy 2 19 the results showed instances of false drops due to partial Topic marginal 23 48 174 homonymy which is very difficultto avoid in Finnish lan­ Search strategy 2 2 11

guage. Partial homonymy means that terms with entirely Incorred term relation­ 3 16 different meanings have some inflected formsthat coin- ships In thesaurus

Tnt. Classif. 17 (1990) No.2 - Kristcnscn/Iiirvclin: Thesaurus in free-text searching 81 remedy to these problems is a linquistically intelligent, mind. This is a simplistic approach especially for abstract context sensitive basic form recognition program either concepts. Concept recognition procedures described in in the storage or in the search phase. [3] may provide an approach to augmenting searching thesauri in providing assistance for searching abstract 4. Discussion concepts. Another issue is related to the users and updat­ We have tested the effectiveness of a searching the­ ing of searching thesauri: should they be tools for a sam'us in free-text searching of a Finnish full-text data­ general audience, a professional group, or should they be base containing newspaper articles. The purpose of such personalized, user specific semantic tools? In the latter a thesaurus is a) to improve search recall by giving sy­ case, the users themselves would be responsible fo rupdat­ nonyms and other alternative terms for the original ing their personal thesauri which therefore might differ search terms (or at least to relieve the searcher's burden in even among persons working in the same organization recognizing and keying in all possible synonyms) and b) and using the same database. to enable systematic broadening and narrowing of a The user interface to the thesaurus needs also consider­ search by presenting conceptual relationships. ation. In this study, the researcher simulated, in effect, The test results indicate (see Tables I and 2) an essen­ the operation of an automatic tool aiding the journalists tial and statistically significant improvement in recall in searching. When such a thesaurus is provided to the with neglible but statistically significantdecline in preci­ users as a routine semantic tool, it may be available upon sion due to the test thesaurus. This is especially true ofthe request, automatically (i.e. queries are extended auto­ performance of the synonym searches. The findingssug­ matically through the thesaurus) or semiautomatically gest that augmenting original queries by synonyms pro­ (i.e. the thesaurus suggests fu rther terms but the user vided by a searching thesaurus yields almost twice as must approve or reject them). many relevant documents as the original queries with a Although Finnish is a highly inflectional language our neglible (10%) decline in precision. The effect of the re­ results are not language dependent is this respect. We lated term searches is more questionable but, if high re­ have considered semantic, not morphological terms rela­ call is necessary, an important possibility. These possi­ tionships. Even the searches were conducted in an Eng­ bilities may not be relevant to many needs of journalists lish-like manner due to the use ofthe morphological pro­ because they may often be satisfied with only a few ar­ grams yielding basic term forms in the construction of ticles. Nevertheless, they are essential possibilities of the basic index of the database. search modification in general, e.g. for different kinds of clientele. 5.- Conclusions The analysis of the net effects of the query extension We have tested the effectiveness of a searching the­ types (Table 3) corroborate these conclusions. The sy­ saurus in free-text searching of a Finnish fu ll-text data­ nonym extension retrieves an essential share of the rele­ base containing newspaper articles. The test thesaurus vant articles with a precision figure which is often toler­ improved search recall by giving synonyms and other re­ able. The related term extension still adds nearly one fifth lated terms for the original search terms. The synonyms of the relevant articles but now the precision is rather given by the thesaurus improved the relative search re­ poor (13%). This means that the related term extension is call, on the average, from 45% to 82% while augmenting worthwhile only when high recall is important. The the searches also by the related terms given by the the­ figures also suggest that the output should be ranked by saUl'US further improved the relative search recall, on the extension type. average, from 82% to 100% (the maximal relative recall The analysis of the false drops show that word of all the searches). The average precision values were for proximity operators and word frequence thresholds the modes one, two and three were 51 %, 41 % and 33%, could be used to reduce irrelevant results due to search respectively. These findings were statistically significant terms not being related in the articles or due to marginal and are not bound to the properties of the Finnish lan­ treatment of the search terms in the articles. Partial ho­ guage. monymy in inflectional languages like Finnish is a diffi­ The substitution of synonyms given by the thesaurus cult problem to solve but does not seem to cause many to the original search terms shows a considerable in­ false drops. crease in recall with a negligibledecline in precision. Fur­ Untested possibilities in our study were: extensive use ther query augmentation by related terms seems appro­ of proximity operations, word frequency thresholds, priate, if high recall is required. The analysis of the netef­ hierarchical (downward) expansion of abstract terms to fects of the query extension types showed an essential im­ cover a broad class of terms, and the efficiencyof scalable provement in recall (nearly 40% of all relevant articles) synonymity. The last possibility means attachingto each with a relatively high precision (31 %) for the synonym ex­ synonym relation an index (e.g. in the range 0.0... 1.0) de­ tension. The related term extension showed further im­ scribing the degree of the synonymity relation. This provement in recall (nearly one fifth of all relevant ar­ could be used in the search phase by allowing only suchsy­ ticles) with a relatively poor precision (13%). These find­ nonyms of original search terms which have a sy­ ings suggest that the search results be ranked by the nonymity relation to the original term exceeding a given search extension type. limit. A study testing these possibilities is under way. In conclusion, searching thesauri seem to be effective The type of a searching thesaurus we have used gives tools for improving search results in free-text searching terms and simple phrases for the terms the user has in his of full-text databases. The findings and the untested

82 Inl. Classif. 17 (1990) No.2 - Kristcnscn/Jarvclin: Thesaurus in frcc�text searching possibilities for search development through searching [22] Salton. G.: Another look at automatic text�retrieval sys­ thesauri suggest the fo llowing topics to be elaborated in lems. Comm.ACM 29(1986)No. 7, p.648-656. [23] Salton, G., McGill, M.J.: Introduction to modern informa­ fu rther studies: searching thesauri for different types for tion Retrieval. New York,NY: McGraw-Hil1 1983. 448 p. full-text databases and different user populations, exten­ [24] Siegel, S., Castellan, N. 1.: Nonparametric statistics for the sive use of proximity operations, word frequency tre­ behavioral sciences. New York, NY: McGraw-Hill 1988. [25JSoergel, D.: Organizing Information: Principles of database sholds, hierarchical expansion of abstract terms to cover and retrieval systems. New York, NY: Academic Press a broad class of terms, the efficiency of scalable sy­ 1 985. 450 p. nonymity relations, and concept recognition procedures [26J Swanson, D.R.: Searching natural languagc by computer. Science 132(1 960)No.3424, p.1099-1 104. for text processing. [27] Swanson, D.R.: Historical Note: Information Retrieval and the Future of an Illusion. J.Amer.Soc.Inform.Sci. Reference 39(l988)No.4, p.92-98. [28] Svenonius, E.: Unanswered questions in the design of con­ [1] Blair, D.C.: Full text retrieval: Evaluation and implications. trollcd vocabularies. J .Amer.Soc.Inform.Sci. 37(1986) Int. Classif. 13 (1986) No.1, p. 18-23. No.4, p.33 I -340. [2] Blair, D.C., Maron, M.E.: An evaluation of retrieval effec­ [29] Tenopir, c.: Full text databases. In : Williams, M.E. (Ed.): tiveness for a full-text document retrieval system. Ann.Rev.Inform.Sci. & Techno!., Vol 19. White Plains, Comm.ACM 28(1985)No.3, p.289-299. NY: Knowledge Industry Pub!. 1984, p.215-246. [3] Croft, W.R: User-Specified Domain Knowledge for Docu­ [30]Tenopir, C.: Full text database retrieval performance. On­ ment Retrieval. In: Rabitti, F. (Ed.): ACM Conf. Res.& De­ line Rev. 9(1985)2, p.149 -164. velopm.in Inform. Retrieval, Pisa, Italy, Sept.8-10,1986, p.201-206. [4J Directory of Online Databases 9(1988)No.1. New York, NY: Cuadra-Elsevier 1988. APPENDIX [5] Dubois, C.P.R.: Free text vs controlled vocabulary: a reas­ Sample Extracts from the Te st Searching Thesaurus sessment. Online Rev. 1 1(1987)NoA, p.243-253. [6] Fugman, R.: The complementarity of natural and indexing The extracts below are not accurate translations fron languages. Int. Classif. 9(l982)No.3, p.140-144. the Finnish test thesaurus. Rather, they illustrate the test [7] Hersey, D.F., Foster, W.R., Stalder, B.W., Carlson, W.T.: Free text word retrieval and scientists indexing: Perform­ thesaurus. Differences in synonyms and their number be­ ance, profiles and costs. 1. Document. 27(1971)No.3, tween Finnish and English are taken into account. The p.167-183. translations given below may not always match most [8J Henzler, R.G.: Free or controlled vocabularies: Some statis­ tical user-oriented evaluations of biomedical information common terms and phrases in English newspapers. In systems. Jnt. Classif. 5(1 987)No. I, p.21-26. many cases, Finnish journalists use both Finnish terms [9J Iivoncn, ly!".: Indeksointituloksen riippuvuus indeksoin­ and terms borrowed from English for a single concept in tiymperistUste [The dependence of indexing results on thc indexing environment]. Tampere, Finland: Univ.of Tam­ their texts. Therefore the extracts below are not as rich perc, Dept. ofLibr. & Inform. Sci., Report 26, 1989. 200 p. with synonyms as the Finnish version of the thesaurus is. [10 ISO 2788. Documentation - Guidelines for the Establish­ The thesaurus was restricted to economic articles of ment and Development of Monolingual Thesauri. Paris: Int. OrgJor Standardisation 1986. 65 p. the newspaper and followed the literary warrant prin­ [ll]Jeppinen, H., Ylilammi, M.: Associative model of morpho­ ciple. logical analysis: An empirical inquiry. Comput. Linguistics The thesaurus relations are reciprocal although this is 12(1986)No.4, p.257-272. [12]Kimberley, R. (Ed.): Integrating text with non-text. Proc. not shown in the extracts. Inst.Inform.Scientists Text Retrieval '85 Conf. London, UK: Taylor Graham 1986. 120 p. Legend [l3] Kimberley, R. Hamilton, C.D., Smith, C.H. (Eds.): Text Re­ SYN synonymous terms trieval in Context. Proc.Inst.Inform. Scientists Text Retrie­ BT broader terms val '84 Conference. London, UK: Taylor Graham 1985. NT narrower terms 152p. .. RT related terms [14] Lehtola, A., Honkela, T., HyUtyniemi, H. and leppinen, SEE see the desrciption at the referred main term H.: Task Oriented Knowledge Representation Languages entry for NLP-Systems. In: Ras, Z. W., Saitta, L. (Eds.): Metho­ Finnish terms with no known English equi­ dologies for Intelligent Systems 3. Proc. 3rd Int. Sympo­ valcnts sium on Methodologies for Intelligent Systems. Turin, italy, 12-15 Oct., 1988. New York, NY: North Holland SAMPLE EXTRACTS 1988. p.250-259. [l 5] Lancaster, F.W.: Vocabulary Control for Information Re­ annual report trieval. Washington: Inform. Resources Press 1972. SYN annual review, annual overview, activity re� [16] Lancaster, F.W.: Vocabulary Control for Information Re­ port, ... trieval. (2nd cd.) Arlington, VA: Inform.Resources Press RT profit, balance sheet, 1986. [17] McDonald, M.: BASIS-Innovation in Data Management. annual review In: Chen, C.C.,Hernon, P. (Eds.): Numeric Databases. Nor­ SEE annual report wood: Ablex 1984. p.219-236. [18] Meunier, J .-G., Bertralld-Gastaldy, S., Lebel, H.: A call fo r atomic energy enhanced representation of content as a means ofimproving SEE nuclear energy online Full-Text Retrieval. Int. Classif. 14(1987)No. I, p.2- lO. atomic energy production [19] Perez, E.: Text enhancement: controlled vocabulary vs. free SYN nuclear energy production text. Spee.Libr.s 73(1982)No.3, p. 183-192. RT nuclear energy, nuclear powcr, atomic en­ [20] Piternick, A.: Searching vocabularies: A developing ca­ ergy, atomic power, energy, nuclear power tegory of online searching tools. Online Review plants,nuclear power stations 8(1984)No.5, p.441-449. [21]Ro, J.S.: An cvaluation of the applicability of ranking algo­ automatic data processing rithms to improve the effcctiveness of full-text retrieval: 1. SYN EDP, electronic data processing Onthe Effectiveness ofFull- Text Retrieval. 1. Amer.Soc. ln­ RT computer, computing, microcomputer, in­ form.Sci.39(l988)No.2, p.73-78 . formation technology lnt. Classif. 17 (1990) No.2 -Kristensen/Jarvclin: Thesaurus in fr ee-text searching 83 balance sheet nuclear energy SYN profit, ... SYN nuclear power, atomicenergy , atomic power RT annual report, annual review, annual over­ BT energy view, activity report, ... RT nuclear power plants, nuclear power sta­ tions, nuclear energy production, atomic en­ fa ctory ergy production SYN industrial establishment, mill, production stablishment paper ind uslry RT industry SYN paper mills BT chemical forest industry import RT paper, cellulose, carton, cardboard BT trade RT cast-trade, soviet-trade, west-trade, eec­ textile trade, roreign trade, export, order, shipment SYN apparel, suit, clothes, clothing RT apparel industry, shoe industry, textile in­ industry dustry NT apparel industry, clothing industry, docky­ ard industry, forest industry, process indus­ textile industry try, ship industry, shoe industry, textile in­ SYN apparel industry, clothing industry, ... dustry, export industry, ... BT industry RT company, factory, firm, industrial establish­ RT apparel, suit, clothes, clothing, textile ment, mill, ... trade information technology NT bilateral trade, international trade, compen­ RT EDP, automatic data processing, electronic sation trade, world trade, foreign trade, im­ data processing, computer, computing, port, export... microcomputer RT markets, marketing, purchases, sales, or­ ders, shipments markets NT home markets, world market, internal mar­ kets, foreign markets, export markets, com­ Address: Prof.Kalervo Jarvelin and Ms. Jaana Kristensen mon markets Department of Library and Information Science. University of RT trade, marketing, purchases, sales Tampere, P.O.Box 607, SF�33 101 Tampere, Finland

INTERNATIONAL CONCEPTUAL ENCYCWPEDIA FORTHE SOCIAL SCIENCES Volume I

INTERCOCTA GLOSSARY CONCEPTS & TERMS USED IN ETHNICITY RESEARCH

Edited by Fred W. Riggs

Here is the prototype of a new glossary concept'ion and construction us'ing the socalled onomantic approach. Some 250 concepts found in the current literature are arranged systematically with their defin'itions, occurrence sources and differing modes of expression. They are made ac­ cessible by different indexes. 1988. 205 p. , OM 48.-

INDEKS-Verlag, Woogstr.36a, D-6000 Frankfurt 50, Tel .: (069) 52 36 90

TERMINOLOGY and KNOWLEDGE ENGINEERING

PROCEEDINGS OF THE INTERNATIONAL CONGRESS ON TERMINO­ LOG Y AN� K NOWLED GE E N GINE E R lNG, 29 Sept. - 1 Oct., 1987, University of Tr'ier, FRG 2 volumes Now reduced ! edi ted by Hans Czap, Trier, and Christian Galinski, Wien OM 75.- both ! 448+256p., DM� (bound), IS8N 3-88672-202-3 and -203-1

The Congress was organized by the International Information Centre for Ter­ minology (INFOTERM) , Wien and the Association for Termi nology and Knowledge Transfer, Trier under the patronage of the Unesco.

INOEKS VERLAG 1987/1988

Order now!

INOEKS VERLAG Woogstr. 36a D-6000 Frankfurt 50 Tel .: 069 - 52 36 90

84 Int. ClassiC 17 (1990) No.2 - KristcnscnjJarvelin: Thesaurus in fr ee-text searching Peter Furniss

A Proposed Methodology for Examining the Provision of Subject Access in the OPAC

Furniss,P.: A proposed methodology for examining the provision more like those created in indexing for existing online bib­ of subject access in the OPAC . liographic databases, and with the use of traditional Int.Classif. 17(1990)No.2, p.85-89, 22 refs. classification in the new environment (9, 10). The development of the OPAC has stimulated renewed interest One of the prerequisites for attempting a radical reap­ in the provision ofsubject access in the catalogue. Significantdif­ ferenccs between the new and earlier formats have prompted the praisal would be to decide on a methodology. Oneinvesti­ suggestion that there should be a radical reappraisal ofthe provi­ gation which involves detailed analysis of how OPACS sion made. A methodology for such a reappraisal is proposed, are constructed and used is in progress at the Polytechnic based on the proposition that a radical reappraisal will need to refer back to the theory on which subject access is based. The of Central London. This research is noted as one of the Five Axiom Theory developed by R.Fugmann is outlined, most significant projects in TEDD's review of recent de­ which is considered to provide a possible basis for the reapprai­ velopments (! I). The purpose of this article is to suggest sal. Some existing data derived from a cooperative cataloguing environment is described which is considered to be relevant to another, complementary rather than alternative ap­ the axiom in Fugmann's theory which is concerned in part with proach, and to use existing data to indicate the kind of re­ the significance of consistency in indexing. The data is analysed sults it might produce. in terms of consistency, with the objective of indicating what kind of results might be obtained from the application ofthe pro­ posed methodology. The conclusion drawn from the results of 2. Theoretical Basis for the Proposed Methodology the analysis is that the methodology may well be worth fu rther exploration. Such exploration will need to examine a morecxten­ The rationale for the proposed methodology is the sive data sample than the limited one used for the analysis under­ proposition that a reappraisal, if it is to be truly radical, taken in the article, and will need also to examine the role other will need to refer back to theoretical principles, on which axioms of the theory might play in implementing a reappraisal ofsubject access. (Author) the revised techniques should be based. The rationale, therefore, necessitates the identification of such prin­ ciples. One assumption is made: that some form of index­ ing will continue to be necessary for a mechanism which provides access to the subject content of a collection con­ 1. Introduction taining a significant proportion of documents in mono­ In his survey of OPACS published in 1985, MAT­ graphic form. Given this, the principles sought have to THEWS reports on 48 operational systems (1). HUD­ address the problem of how the searcher, when devising SON discusses the impact made on the cataloguing oper­ search strategies, can predict what assignments have ation of six institutions which have adopted the new for­ been made. mat. Considerable benefits are reported from its adop­ A number of theoretical approaches have been sug­ tion, for both the production and utilisation of the cata­ gested, but none yet fully developed and evaluated. Four logue, and these are expected to promote its use by a grow­ were discussed by BORKO in 1977 (12), and two are ing number oflibraries (2). noted as significant in the context of retrieval effective­ A number of writers discussing the impact of the new ness by ROBERTSON (13). The techniques discussed by fo rmat, especially on subject access, have asserted that its both authors use computational methods to calculate adoption provides the opportunity, perhaps the utility and probability. SALTON, one ofthe theorists dis­ necessity, for a radical reappraisal of the provision made cussed, concludes in a later article that while much has in earlier formats (3-6). been achieved using this methodology, much remains to WILLIAMSON, reviewing subject access in the on­ be done in developing this line of approach (14). More­ line environment in 1984, commented that much ofthe re­ over, the methodology has mainly been tested on types of search up to that point had been devoted to adapting documents amenable to fu ll text searching, such as peri­ traditional techniques to the online environment, rather odical articles, or abstracts of these. An approach sug­ than attempting a radical reappraisal (7). A later com­ gested by COOPER does appear to be equally applicable mentator observes that much development has been to monographic material, bus some doubt remains as to determined by commercial viability, and not by the need the suitability ofthe methodology to such material (15). for a fu ndamental reappraisal (8). Two projects which An alternative approach has been suggested by FUG­ exemplify the adaption of existing methodology to the MANN, who specifically mentions the revision of exist­ new fo rmat have been concerned with increasing the ing systems as one possible area of application for his the­ depth of indexing in order to make searchable records ory. He developed a Five Axiom Theory for Indexing

Int. ClassiC 17 (1990) No.2 - Furniss: Subject access in the OPAC 85 and Information Supply, which appears to be less de­ records supplied by the same agency, in the case ofdocu­ pendent on computation than methods utilising utility ments for which both had created records. and probability theory. One article in which the theory is The data was collected, therefore, to determine the ex­ explained cites 45 references which include several con­ tent to which indexing assigned by two agencies agreed, tributions based on such methods, but cites earlier work in cases where both had created records. This agreement by RANGANATHAN and RUSH and LANDRY as was considered to be a readily available and reasonably precursors of the Five Axiom Theory (16). The theory objective indication of the degree of consistency a cooper­ has been developed in the context of chemical documen­ ative member taking records from both agencies might tation, and a need is acknowledged for testing its relev­ expect to find in the descriptors assigned by the agencies ance in other areas. One area of potential application to the documents it acquired. could be the OPAC, which, because of its role in docu­ menting monographic material, relies heavily on index­ Indexing consistency is one of the topics considered by ing for subject access. Fugmann in his discussion of the Five Axiom Theory. The discussion, mainly in terms of predictability and fi· One of the stated constraints which has prevented full delity of indexing, results in the conclusion that indexing testing of the theory in operational systems is the lack of consistency is not a useful overall criterion of indexing time available to workers in such systems for testing new quality. Instead, representational predictability is pro­ theories (17). Some relaxation of this constraint wouldre­ posed as such a criterion. The discussion is in terms of a suIt if existing data could be used to test the theory. The two stage indexing process: selection of concepts to be in­ following section explains the derivation of some data dexed and choice of descriptors to represent these con­ which appears to be relevant to part of Fugmann's the­ cepts. At the concept selection stage, indexing consist­ ory, and therefore suitable for a preliminary exploration ency is considered to be conducive to, but not sufficient of the kind of results which could be derived from using to ensure, the predictability of concepts selected to index the theory as a basis for a reappraisal of indexing needs. a document. At the representation stage, consistency in Such a preliminary testing of the viability of the methodo­ the choice of descriptors to represent the concepts se­ logy is advisable in view ofthe investment of time and ef­ lected for indexing is considered neither necessary nor fort required for full testing. sufficient to ensure predictability. Consistency then is not seen as having a central role in 3. Origin and Relevance of the Data Utilized ensuring representational predictability, only as one fac­ The data in question was collected during the initial tor promoting its attainment at onc ofa two stage index­ stages of the automation of the cataloguing operation ing process. Analysis of data on indexing consistency will from which it was derived, when an inhouse system was therefore be in terms of a peripheral part of the Five abandoned in favour of membership of a cataloguing co­ Axiom Theory. Such an analysis can nevertheless have operative. As a part of the change, Library of Congress value at a preliminary stage, where the objective is an in­ subject headings were substituted for locally created itial test of a proposed methodology, by using it to ana­ headings in the subject heading catalogue. At the time lyse some data, in order to determine what kind of results the data was collected, evidence was sought on the effect are obtained. Any part of the theory, even a peripheral of adopting headings assigned by the two major catalo­ part, can be used at this preliminary stage, though guing agencies which supplied records to the coopera­ plainly, if results prove encouraging, much will still re­ tive, without local modification, or local selection when main to be done in exploring the role the whole theory several headings were assigned to a document. Heading might play in developing the methodology further. In the assignment was done locally by professional staff, and following sections, further details of the data are given, was thought to be one of the areas in which automation and an analysis of the kind outlined above is made in might reduce the amount of staff time required. One of terms of indexing consistency, one of the factors con­ the factors identified as determining the effect of aban· sideredin the development of the theory on which the pro­ doning local editing of heading assignments was the COil­ posed methodology might be based. sistency with which headings were assigned by the agen­ cies, and the extent to which there was agreement be­ 4. Deseription of the Data tween them on headings assigned. The data collected were the subject descriptors as­ Agreement between the agencies was considered to be signed by the British Library and the Library of Con­ the easier to measure objectively, since for documents gress, and in the local catalogue, to documents added to catalogued by both, a direct comparison of descriptors the local collection between 1978 and 1980 in the area of assigned could be made, and this comparison indicates librarianship and information science. Two kinds of de­ the likely extent of agreement between descriptors as­ scriptos were assigned to MARC records by both agen­ signed to a topic in documents for which records from cies and also used in the local catalogue: Dewey Decimal one only of the agencies are available. Classification class numbers and subject headings fr om The indexing of documents for which a record from the Library of Congress Subject Heading List. Table I only one agency was available was thought to be presents the basic fa cts on the size of the data sample, and specially significant in determining indexing agreement, the extent to which records were available from both since any differences arising fr om differing agency index­ agencies. Agency names are abbreviated to BL and LC in ing policies could be eliminated by always selccting the this and subsequent tables.

86 lnt. Classif. 17 (1990) No.2 - Furniss: Subject access in the OPAC Table I: Total documents added to the local collection by non-topical qualifiers, such as form or geographical treatment. BL and LC record found 264 BL record only 87 Table 3: Analysis of class number differences LC record only III No agency record 134 Different main class numbers 35 Third digit differences 37 Total 596 Post decimal point differences 65

5. Analysis of Data Total 137 Table 2 analyses the descriptor assignments to docu­ ments catalogued by both agencies. The table assumes Table 4 contains the result ofa similar qualitative ana­ one class number per document. In cases where dass lysis of the subject heading assignment data ofTable 2. It numbers from more than one edition of the Dewey Deci­ was based partly on the structure of the Library of Con­ mal Classification were assigned, that closest to the 16th gress Subject Headings, partly on the way such headings edition, from which most of the class numbers in use in are used to index the subject content of documents. The the local catalogue are derived, was recorded. The sub­ structural element was based on the use of subdivisions ject heading totals record the number of times the same to designatespecific aspects of a heading. Many of the dif­ subject heading was assigned by both agencies to the ferences noted consisted in the use of such subdivisions. same document, and the number of times a heading was Such differences are considered analogous to the kind of assigned by one agency only. Exactly matching descrip­ difference included in row 3 of Table 3 in the class num­ tors only are counted as assignments by both agencies. ber analysis. On this basis, Table 2 indicates that the agencies difTer considerably in descriptor assignment. Analysis of differences other than of subdivision was based on the way both agencies use Library of Congress Table 2: Analysis of documents catalogued by both agen­ Subject Headings to index the subject content of docu­ cies ments. Both frequently assign several headings to a docu­ ment, which can result in partial agreement in the assign­ Documents catalogued by both 264 ments made. The differences were divided into two ca­ Same class number assigned by both 127 tegories: those derived from documents on which assign­ Headings assigned by both 230 ments were partially agreed, and those derived from do­ Headings assigned by one only 395 cuments on which there was no agreement. The first ca­ tegory is identified in Table 4 as headings additional to The analysis in Table 2 is based on the extent to which agreed assignments, and is considered analogous to the descriptors in the sample match exactly. In assessing the kinds of differences included in rows 2 and 3 ofTable 3, in extent of agency agreement on assignments a less rigo­ that there is some measure of agreement between the rous definition may be advisable. There is, for example, a agencies. The second category, headings assigned to do­ significant difference between disparate class numbers cuments with no agreed headings, is considered analo­ which repr.esent different decisions on the main focus of a gous to the more radical class number differences of row document, and class numbers which differ merely by the 1 of Table 3, in that there is no element of agreement be­ addition of a form or geographical extension to an tween the agencies on descriptor assignment. agreed concept designator. Differences of the latter kind are readily taken into account by truncation of the more Table 4: Analysis of subject heading differences specific class number. Differences of the former kind may require a complex disjunctive search statement, if all Headings with differences of subdivision 124 possible variations are to be included. Headings assigned in addition to agreed headings 165 Table 3 represents the result of an attempt to take such Assignments to documents with no agreed headings 106 qualitative differences into account in further analysing the class number divergence shown in Table 2. The ana­ Total 395 lysis is based on the structure of the Dewey Decimal Classification. The assumption, generally valid, is made Tables 2-4 indicate that even if agreement between that the earlier the divergence occurs between two class agencies is defined less rigorously than the assignment of numbers read left to right, the more radical is likely to exactly matching descriptors, considerable differences in have been the difference between the conceptual analysis assignment remain. 52% of documents are assigned dif­ of the main focus of a document. ferent class numbers, and of these, 25% are differences of The table concentrates on three significant points in main class number, suggesting different conceptual ana­ the divergence continuum: differences in main class num­ lysis ofthe main focus of a document. 63 % of all heading bers; differences at the third digit, indicating choice of dif­ assignments are made by one agency only, and 170/0 of ferent subdivisions of the main class; differences beyond headings represent radical divergence between the agen� the decimal point, frequently indicating a more specific cies, in that no agreed headings are assigned to a docu­ subdivision of the same topic, or the same topic qualified ment.

lnt. Classif. 17 (1990) No.2 - Furniss: Subject access in the OPAC 87 Assignments of different main class numbers or totally Table 5. Documents ass igned different main class numbers 1 016.02100973 021 .00973 American library history:a different sets of subject headings to documents appear to bibliography offer the strongest evidence of differences in agency ana­ 2 016.0253 025.3 Indexing and abstractlng:an lysis of the subject content of documents. Such descrip­ international bibliography 3 016.0255 020.7 Library Instruction:a tors need not be wholly inconsistent, however. Dif­ blbl iography ferences could result from consistent choice between op­ 4 016.091 091.094 Medieval scribes, manuscripts and libraries tions allowed by the classification scheme, or between 5 658.154 029.11 Zero based budgeting In hcadings related by the syndetic structure of the subject I ibrary management heading list. Such consistent, and therefore predictable, 6 658.1594 070.5 Scholarly publlsh('r's guide: choices would reduce the element of uncertainty to the financial and legal asp('cts user of an operational system. A fu rther analysis of the 7 658.401 025.1 Planning approach to library rnanag('rnent subset of documents with the most disparate descriptor 8 658.4032 025.1 Syst('ms analysis in· libraries assignments was therefore made, to establish whether 9 65£1.45 025.1 Commun ication in library there is any evidence of such underlying consistency. manag",ment 10 656.809070573 070.57 Information through the Table 5, detailing items assigned different main class printed word numbers, is arranged by class numbers other than those 11 658.8090705730947 338.4707057 Soviet book publishing policy 30947 from Dewey Decimal Classification main class 020. The 12 658.91021 025.1 Library structures and first four rows may bc considered to exemplify the kind staffing systems ofpredictable choice for which evidence was sought. The 13 658.9102100 027.742547 Loughborough Library Management Unit report to choice between grouping topical bibliographies with the the British Library topic or at the bibliography number is probably one of 14 658.9107050941 070.50941 The small publisher:manuai and case histories the Dewey Decimal Classification options which can be 15 001 .5 029.7 Implementation of complex most readily recognised and consistently applied. Rows information systems 5 to 14 exhibit a discernible pattern: the choice between a 16 001 .552024092 025.17 Printed ephemara 17 001 .64404 029 Formal models and practical business technique or operation and its employment in tools for information systems the field of library administration or publishing. Rows design 18 00 1.6443 621 .3819534 Guide to computer output 15 to 18 exhibit a similar pattern of choice between alter­ microfilm natives: the provision at 001 for the general process of 19 011.2 025.060013 Reference work In the communication, and separate provision for systems and humanities 20 070.5 025.236 LibrHians Boo�sellers and materials employed in that process. The range of class Publfsh('rs Working Party: numbers in the second column, however, does not sug­ LAIG guidelines 21 070.594 029.9001430973 Enquiry into scholarly gest the kind of consistent choice between alternatives communIcation which could easily be predicted by searchers, and this is 22 300.141 070.572 Scientific publication system in social sciences certainly true of the remainder of class numbers in the 23 302.23 070.502854 future of the printed word table. 24 338.47027473 n27.473 Public Ifbrarfes:an economIc The analysis of totally disparate subject heading sets is View based on the relationships between headings. Two rela­ 25 338.768620942951 070.90942951 History of th., Gregynog Press tionships common to many thesauri are used: broader­ 26 346 .410482 070.52 Publishing agreements 27 351 .004094 025.49351004094 L i st of uniform headings fDr narrower term, and related term. Most relationships be­ higher leglslaUv(' and ministerial bodies in twcen headings of these two kinds will fa ll within two European countries broader kinds of relationship used in the analysis of in­ 28 352.16 029.9352 Local government and dexing languages: syntagmatic and paradigmatic. Information technology 29 501.41 029.95 Commun icaUon:the essence of The introduction of these broader relationships into science 30 574.072041 029.9574 Biological Information Review the field of information retrieval from that of linguistics Committee report has been attributed to Gardin by Foskett. In the original 31 610.7 026.61 How to use a medical library usage, a particular relationship could be of either kind, 32 686.43 025. 179 Serials management and depending on the indexing language used (18). Various microforms definitions of the terms have subsequently been given (19 33 727.8 022.3 Mason on library bu ildings 21). Common to most is the idea that paradigmatic rela­ 34 745.67 091 The illuminated manuscript 35 973.07041 026.773 American Studies Library tionships are of a more permanent nature, deriving from Conference the definitions ofterms, rather than the more transient re­ lationships between terms found in documents. Most broader term-narrower term relationships will be of the ment of differing policies on indexing specificity. Specific paradigmatic kind, related terms are more likely to fall descriptors may equate to a more general term, linked by within the syntagmatic category. the paradigmatic syntactic structure of the indexing lan­ The two general kinds ofrelationship between descrip­ guage, used at a less specific level of indexing. This is a tors may be significant inexplaining the choice of differ­ less likely explanation of indexing diffe rences where the ent descriptors for documents, If the relationship be­ relationship between selected terms is syntagmatic, since tween selected terms is paradigmatic, an agreed assess­ there is no such general related term which represents an ment of the subject content of a document may result in agreed broad analysis of the subject content of a docu­ the choice of different descriptors owing to the employ- ment.

88 Int. Classif. 17 (1990) No.2 - Furniss: Subject access in the OPAC In addition, paradigmatic relationships may be margi­ dexing was to be taken uncritically fr om available rec­ nally easier to incorporate in searching, perhaps because ords. The problems presented by this kind of consistency of their permanent nature. This is most clearly evident in do not appear to be ones which would be solved by in­ the case of generic-specific relationships: moderation of creasing indexing depth, nor by exploiting the more flex­ a strategy to incorporate more or less specific terms is a ible approach to subject access made possible by the likely early response to an inappropriate level of recall. OPAC, in an attempt to utilise techniques developed in Table 6, appended to the end of the article, analyses to­ online bibliographic searching. tally disparate heading sets in terms of paradigmatic and Alternative approaches which take account of index­ syntagmatic relationships. Columns 2 and 4 1ist the head­ ing theory need to be considered, if a truly radical reap� ings assigned by each agency to a document. Column 3 in­ praisal of subject access in the new OPAC environment is dicates which of the column 4 headings are related to col­ to be realised. umn 2 headings in the 11th edition of Library of Con­ A methodology for implementing such an alternative gress Subject Headings (22). P indicates a column 43 approach has been briefly examined, using an available heading which is designated a broader or narrower term limited sample of data, and a small part of a significant under one of the headings in column 2. S indicates a re­ contribution to indexing theory. The results suggest that lated term. A relationship is classed as P if the column 4 the methodology is worth fu rther development, to ex­ heading is treated in LCSH as a broader or narrower ploit fully the whole of the theory used for the prelimi­ term of a heading not assigned by either agency, which is nary exploration reported. related to a column 2 heading in either way. Since Li­ brary of Congress Subject Headings are modifiedcontin­ References ually, some additional headings may have been related at (1) Matthews, J.R.: Public access to online catalogs. 2nd ed. the time the assignments were made. The list is relatively Neal-Schuman 1985. stable, however, and the number of related headings (2) Hudson, J.: Cataloging for the local online system. Inform. Techno1.& Librs. 5(1986)No. I, p.5-27 shown in table 6 is unlikely to be significantly increased (3) Bates, MJ.: Subject access in online catalogs: a design by such headings. In most cases, several headings are as­ model. J .Amer.Soc.lnform.Sei. 37(1986)No.6, p.357-376 signed by both agencies to a document, and relationships (4) Hildreth, C.R.: Beyond Boolean: designing for the next generation of online catalogs. Libr.Trends 35(I987)NoA, between such mUltiple headings can be complex. The at­ p.647-667 tributions in column 3 are not intended to map such com­ (5) Lewis, D. W.: Research on the use of online catalogs and its plex relationships, merely to indicate that a column 4 implications for library practice. 1.Acad.Librnship 13 (1987) No. 3, p. t52-157 heading fa lls within the category in relation to at least (6) Lynch, c.A.: Cataloging practices and the online catalog. one heading in column 2. The complex structure of Li­ In: ASIS'85: ProcA8th ASIS Annual Meeting. Las Vegas, brary of Congress Subject Headings also requires a some­ 1985. Koowledge Industry Pnb1. t 985. p.11t -t t 5 (7) Williamson, N.J.: Subject access in the online environment. what arbitrary definition of what constitutes a totally dif­ In: Advances in Librnship 13(1984)p.49-97 ferent heading. Included in Table 6 are phrase headings (8) Chitty, A.B.: Indexing fo r the online catalog. Inform.Tech­ where some element of the phrase is different, for no1.& Librs.6(1987)NoA, p.297-304 (9) Mandel, C.A.: Enriching the library catalog record for sub­ example Document 10 and headings which appear as sub­ ject access. Libr.Resources & Techn.Serv. 29(I985)No.I, headings of other headings, for example Document 2. p.5-t5 Although all except one oftherelationshipsfall into ca­ (lO)Markey, K., Demeyer, A.N.: Findings of the Dewey Deci� mal Classification online project. Inl.Cataiog. 15 (1986) tegory P, a number of headings are not related at all in No. 2, p.15-19 LCSH. Differing policies on indexing specificity cannot (I l)Tedd, L.A.: Computer-based library systems: a review of explain, therefore, all the assignment differences, and the the last twenty-one years. J .Doc. 43(l987)No.2, p. 145-165 (l2)Borko, H.: Toward a theory of indexing. Inform.Process.& evidence fr om subject headings tends to confirm that Management t3(t977)No.6, p.355-365 from class numbers. No pattern of underlying consist­ (13)Robertson, S.E.: Indexing theory and retrieval efTective­ ency is evident which would explain descriptor assign­ ness. Drexel Libr.Quart. 14(1978)No.2, pAD-56 (I4)Saiton, G.: Historical Note: The past thirty years in informa� ments to documents manifesting the greatest variance be­ tion retrieval. J.Amer.Soc.Inform.Sci. 38(1987)No.5, tween the agencies. p.375-380 (I5)Cooper, W.S.: Indexing documents by Gedankenex­ perimentation. J.Amer.Soc.Inform.Sci. 29(19788)No.3, 6. Conclusion p.t07-t 19 (16)Fugmann,R.: The five-axiom theory ofindexing and infor­ The combined evidence from documents with differ­ mation supply. I.Amer.Soc.Inform.Sci. 36(l985)No.2, ent main class numbers assigned by the agencies, or with p.lt6-129 no shared heading assignments, tends to confirm the re­ (17)Fugmann, R. ibid., p. 126 (18)Foskett, A.C.: The subject approach to information. 4th ed. sults fr om the whole sample. Considerable diversity in Bingley 1982. p.90-9 t the descriptors assigned to documents exists. Wile some (19)British Standards Institution. BS 5408:1976. Glossary ofdo� of the divergence could be attributed to differences in spe­ cumentation terms. (20)Buchanan, G.: A glossary of indexing terms. Bingley 1976. cificity,or to non-topical additions to descriptors, signifi­ (21 )Lancaster, F.W.: Vocabulary control for information retrie­ cantly different assignments were made in many cases. val. 2nd ed. Information Resources Press 1986. Little evidence of differing, consistently applied, index­ (22)Library of Congress. Subject Headings. II th cd. Washing­ ton, DC: Libr.of Congr., Cataloging Distribution Service ing policies was evident. The conclusion must be that in­ 1988. dexing consistency, one fa ctor identifiedin Fugmann's theory as conducive to effective indexing, would not bere­ Mr.Peter Furniss, 2, Burnt Stones Grove, Sheffield S 10 alised in the operational environment examined, were in- STU, England

lIlt. ClassiC 17 (1990) No.2 - Furniss: Subject access in the OPAC 89 Table 6. Documents assigned different subject heading sets

Col.l Col.2 Col.3 Co1.4

Acquisition of serial publications Librarles,Special-Great Britain library surveys-Great Britain Libraries,Unlversity and college-Great Britain Periodical selection

Acquisitions (libraries) libraries-University and col lege-United States -Acquisitions

Acqu i s Iti ons (IIbra r i e s) -Addres ses, Collection development (Libraries) essays, lectures

4 Book selection Collection deve lopment (Libraries) Selection of nonbook materials

5 Aged Libraries and the socially handicapped-Great Handicapped Britain Hospital.libraries Libraries and the socially handlcapped­ libraries and the aged United States Libraries and the physically handicapped Library services Prison libraries Prisons

Authors, British Authorsh I p -H I story Copyri g ht-H i story Publ ishing and publishers-History Society of authors-History

Blblographical services Comun ication In science-United States R�search Libraries Commun ication of .technical information-United Scholarly periodicals . States Scholarly publishing University presses

8 Bibliographical services-Case studies Information storage and retrieval systems-Case Machine readable bibl iographic data- studies Case studies

Boo� industries and trade-Collected works Scholarly publishing Libraries-Col lected works L i braries-Schoiarly publl sh lng-Col lected works Scholarly periodicals-Col lected works

10 Catalogs, Union-Great Britain German language_Li brary resources-Great Britain German phl lology-Bi bl iography-Urlion li sts German literature-Library resources-Great Britain

\I Citizens' associations-Great Britain libraries and commmun ity-Great Britain PubI ic I ibraries_Informati on services­ libraries and metropolitan areas-Great Britain Great Britain Libraries and the soc ially handicapped-Great Britain

12 Communication In ;Jhysics-Great Britain PhysIcs-Information services

13 Communication in science Documentation Communication of technical information Information networks Information services

14 Communication in science Information storage and retrieval systems­ Communication of technical information SCience-Handbooks, manuals, etc. Information storage and retrieval systems Technology-Handboo�s, manuals, etc.

15 Data transmi ssion systems Teletext (Data transmission system) Viewdata (Data transmission system)

16 Electron I c data proces sing-Addres s es , Quest i on-an swer i ng systems-Congres ses essays, lectures Information storage and retrieval systems­ Addresses, essays, lectures

Fiction in libraries 17 li braries-Speci a I coil ectl ons-F lction Public libraries-Great Britain collections

Reference services(libraries}-Great Britain 18 Information guides-Great Britain

Referral centres 19 Information networks Information services Reference serv i ces-l i brarl es

Information services-Congresses 20 Information science-Congresses Library science_Congresses

On-line bibliographic searching-Congresses ". Information services-Congresses Information storage and retrieval systems -Cong resses On- line data processing-Congresses

Subject headings-Information science 22 Information storage and retrieval systems

0 library resources-Great Britain-Directori s 23 lib ra r i es -Grea t B rita i n- i rec tor i e s � Research libraries-Great Britain-DirectorIes

Library education-Great Britain 24 Libraries and readers libraries, University and college LIbrary orientation-Great Britain

Library fines and fees-Great Britain-Congresses 25 Library finance-Great Britain-Congresses Photocopy services in libraries-Great Pub I ic libraries-Great Britain-Congresses 8r ita i n-CongreS ses

Public libraries-Great Britain-Administration­ 26 LIbrary sclence-Great Britaln-History- 20th. century-Congresses Addresses, essays, lectures

Scholarly periodicals-Publishing 27 Little magazines-Economic aspects PubI i c librar I e s -Eng I and-Newcas t I e-upon-Tyne 28 tlewcastle-upon Tyne City Libraries­ History History

* Two documents, reporting successive years of a conference, assigned identical headings. Counted as 8 assignments.

90 Int. Classif. 17 (1990) No.2 - Furniss: Subject access in the OPAC Lei Zeng University of Pittsburgh

Establishing a Compatible General Vocabulary in China: the Capability*

Zcng, Lei: Establishing a compatible general vocabulary in Considering these facts in China as described above, China: the capability. this study attempts to findan access to the establishment InLClassif. 17(1990)No.2, p.91-98, II refs. of compatibility! among thesauri, that is, to build a The study attempts to findan access to the establishment ofcom� general, compatible Chinese vocabniary based on the patibility among Chinese thesauri, that is, to build a general com­ existing thesauri in China and abroad. After a general patible Chinese vocabulary based on existing thesauri in China and abroad. After a general analysis of the factors influencing analysis of the factors influencing compatibility between compatibility between thesauri, the compatible capabilities be­ thesauri, the compatible capabilities between special the­ tween special thesauri and general ones, as well as between sauri and general ones, as well as between Chinese the­ Chinese thesauri and English ones are discussed. (Author) sauri and English ones are discussed in this paper.

1, Fa ctors that Influence Compatibility Between The­ sauri: a General Analysis 0, Introduction There are many factors that contribute to the compati­ In the mid-1970's, twenty years after the fo unding of a bility of thesauri. The extent of overlap in subject matter, central controlled nation-wide information service sys­ specificity, and vocabulary size are commonly believed tem, the Chinese governmentlaunched a plan for develo­ to play important roles (5), (2). The degree of pre-coordi­ ping a national infOlmation retrieval system which will in­ nation within terms (6)2 and the extent to which vocabu­ clude not only products of databases from other coun­ laries are 'constructed' (5) have also been shown to in­ tries, but also Chinese resources. The development of in­ fluence compatibility. Emphasis on the structure and the dexing thesauri, as a pilot project, started in 1974 when a quality of a hierarchical display, is also involved (7). 1. general thesaurus, "Chinese Thesaurus", was designed Dahlberg stated that the complexity of any thesaurus, to provide a working vocabulary in all fields of human which is determined by the number of pre-combinations knowledge. This ten-volume thesaurus contains approxi­ of the concepts involved, the manner in which concepts mately 109,000 tenus and covers all subjects organized are described and the structural components of the the­ into 58 broad subject fields. saurus are the factors that cause limitations in the estab­ In the decade 1974 to 1984, approximately 20 thesauri lishment of compatibility between ordering systems (2). devoted to special areas were developed (11). During the To summarize the factors that influencecompatibility of past four years (1985-1989), the number of special the­ thesauri, we may deal with such factors under the fo llow­ sauri has doubled, while more are being planned or are ing aspects as shown in Table 1. under development. As a result, over 20 databases, each (1) Principle aspects. Principles, which have the most of them owns over 10,000 Chinese titles now, have been direct influenceon compatibility, are decided at the first merged into the information retrieval services which step of thesaurus design. They may include at least three were originally based on 53 files introduced from other parts, that is: countries. However, as problems caused by the lack ofin­ (Ja) In itial term-structure principle. A term-structure dexing languages are diminishing, problems of access based on either pre-coordination or post-coordination become dominant. was chosen at the beginning. Meanwhile, the degree of Not only is the development of each thesaurus a labor­ precombined tenus in a post-coordinated vocabulary intensive, ongoing and repetitive job, but also is search­ was also decided. ing in merged files which had been indexed with different (1b) Original orientation of vocabulary design. It is thesauri becoming more and more complicated. Another not difficult to find that quite a few vocabularies pub­ problem is that although a large, general thesaurus is al­ lished in the U.S.A. and Europe were originally designed ways considered as a potential compatible and conver­ for certain collections or projects. However, in China, tible basic vocabulary, the Chinese Thesaurus cannot most thesauri were designed according to the classifica­ take on this role. It is, in fact, merely a reference the­ tion of disciplines, science and technology, while a var­ saurus rather than an actively used one because of its in­ iety of distribution of documents in different collections herent problems, and because it has never been renewed were ignored. Such discipline-oriented vocabularies are or revised since its printing in 1979. likely to be highly different from collection- or role-

Tnt. Classif. 17 (1990) No.2 -Zeng: Compatible General Vocabulary in China 91 One thing that should be noticed besides the factors men­ Ilnlli.1 tioned above is that in any case, the more thesauri are

I �r9!ee o! -(- ion built on the same principles, cover alike vocabularies, Principle - and use the same structure, the greater will be the degree Orlgin61 Co i Principle Aspects 1" i of compatibility between them_ OrientationJ of D"lgn Role-o",,'" Mon",1 relrievol - Init161 Purpose of Feom moo",1 0\0, ( .) Vocabu16ry Usage I , • 2. Aw kward Compatible Capabilities Between Special SIZe - Speclflelly Thesauri and General Thesauri • Vocabulary Term Form.' Conleol 2A. Statement of the problems Aspects • Enlry P01nl • There are some issues describing experimental ap­ I 1of Expr"si" • proaches to establish compatiblity among special the­ Construction DlsPl', • sauri with a high level of subject matter overlap_ The 76% Subjecl legory Displ., · Aspects F.e.' automatic mapping of a sample of MeSH to the Agricul­ • 3 • tural/Biological vocabulary (9) and the rich data fr om Term , DISplay the experimental study of convertibility between ASTIA I lOf Term and AEC Subject Headings (3) show a high compati­

:,_Of Term ' • bility existing between special thesauri with similar sub­ Other Aspects in- ject coverage_ In contrast, the capability of achieving :! Of lerm ( .) in compatibility between a special thesaurus and a general :;,:y or V.gue - one is believed to be very low, see for example the result of less than 11% mapping of LCSH to the Agricultural/Bi­ Table 1: The Factors that Influence Compatibility ological Vocabulary (9)4 _ Remarks: + --- Promoting compatibility - --- Impeding compatibility ( )--- Uncertain 2B. Traits associated with the compatiblity of general thesauri oriented ones in their vocabulary size, specificity, and A general thesaurus has a vocabulary covering several subject coverage. disciplines or subject fields_ It is always discipline­ ( 1 c) In itial pUipose of vocabulary usage_ Although al­ oriented rather than collection-oriented or role­ most all thesauri published in Northern America and Eu­ oriented. Internal contradictions of general thesauri are ' rope were developed for mechanized retrieval, China has commonly recognized as the fo llowing: to take manual retrieval into account because of a severe (1) Inverse relationship �f coverage and specificity_ Al­ shortage of technical facilities and a retarded improving though the vocabulary in a special thesaurus may cross of technology_ Therefore, some Chinese thesauri have a several disciplines, we do not consider it to be a general very high degree of precombined terms_ thesaurus because it covers only closely interrelated disci­ 2. Vo cabulary aspects. Vocabulary aspects are also plines and has only a small term size for each ofthesedisci­ very influential on the conversion among thesauri, but plines. This strategy ensures its rich vocabulary in its they are flexible. Size, specificity, format control, entry special field. It is quite different in a general thesaurus. point, and precision of expression vary among thesauri, First, its vocabulary has to cover as many disciplines as therefore they make the results of conversion between needed, among which there might exist only a loose inter­ any two of such thesauri variable. relationship. Second, it has to provide an average size vo­ 3. Structure aspects. Structure aspects have an indi­ cabulary in every discipline or subject field_ As a result, rect influence on conversion. Most thesauri use extra the vocabulary of each discipline or subject field in a structure to display terms and their relationships with general thesaurus is usually less in number and specificity other terms besides an alphabetical list Hierarchical dis­ than that of a special thesaurus_ Table 2 shows four the­ play and pennuted display are the commonly used ones sauri which cover from fourteen to one majors. When in thesauri published in Western languages, while subject checking the specificity level of them by examining their category display is seen as the most helpful structure in hierarchical trees, (here only three or more level hierarchi­ China and of course is very popular in all Chinese the­ cal displays are considered as "tree"_) we notice that the sauri. Anyway, itis believed that the more a thesaurus dis­ percentage of four or more level trees among all trees in­ plays its vocabulary, the more helpful to conversion. crease as the majors covered by a thesaurus decrease. See 4_ Other aspects which influencecompatibility can be the examples given in Table 2. The percentage increases derived more or less from characteristics of language, from II % to 19_5 %, 14.1 %, and 22_9 % with the de­ e.g. multi-lingual conversion with its many special prob­ cline of the covering majors fr om fourteen to five and two. lems; frequency of term occurrence in the indexing and Since many trees in a general thesaurus have less hierar­ searching process; assignment of index terms; as well as chical divisions than in a special thesaurus, terms in a the presentation of concepts in the thesaurus, and in the special thesaurus have less opportunities to find their indexing and searching stage. These are rather varying equivalents at the conversion or mapping stage as shown items and are influenced by SUbjective fa ctors. inFigA_

92 Int. Classif. 17 (1990) No.2 -Zeng: Compalible General Vocabulary in China 41-lml ofD.("{�ton ma'or I (A+B)/T*1009ll l, Chines, :hlnese 3 4 s TheS6urUS 12,423 2,821 65,200 2 . 9 43 (Sci.& Tech 2 Russian rheS6UrUS of 33 3,533 1,191 14,825 31.9 Sci. & Tech.

3 En91ish & 1'l6tlon61 Chinese Defence Thesaurus 2,727 545 17,173 19.05 5 or ScI. & Tech. 4

Chints@ Aeron6utic Fig A. 455 166 7,505 8.27X Thesaurus(2nd Ed1.) 1

(2) Inverse relationship of a common vocabulary and a Table 4. Percentage of Top-level Terms plus Isolated Terms. special vocabulary. By examining 'F' (see Table 4), we may find that There is also a problem in selecting terms for a general general thesauri usually possess a larger number of top­ thesaurus when its final size is limited. Generally, in level terms and isolated terms (see the results of F ofN 0.1 general thesauri, terms that may be used in thepost-coor­ and No.2 thesauri in Table 4) than special ones (see No.3 dinationprocess in more than one subject field are and No.4). This, too, brings difficulties to the conversion preferred so that the vocabulary may present as many or m�pping between general and special thesauri. concepts as possible and keep the size as small as As shown in Fig B, C, and D, due to the reason dis­ possible. Many special terms are expected to be post-co­ cussed above, part of the terms in a special thesaurus ( ordinated with common terms when necessary. circle 3 ) are likely to be excluded by a general thesaurus Table 3 shows the average number of descriptors of (circle I plus 2). each subject field in four general thesauri of science and Fig B. Vocabulary ofa FigC. Vocabulary ofa technology, each of the thesauri covers hundreds of sub­ general thesaurus special subject field ject fields (see the subdivisionsin Table 3). that is included in a general thesaurus

Broad Numb�r of d�scriptors Ilall!! Subdi- olasslfl Title uage vislolls III a subdivision Possible increasing subjeot 'Tota1 fields Min M.. Ave eaoh subd. 10bl . Chillese 0 5020 T�esaurusV Chi 43 502 65200 2 2250 129.9 (Sol & T�ch) 50 25100

10 1880 1040 TEST '" 22 188 17810 2 94.7 50 9400

10 2080 JICST Jo, 14 20' 24348 117.1 50 10400 FigD. Vocabulary ofa

Thuaurus 10 3020 special subject R" ofSciellce 33 302 14825 4 541 49.1 field that is & Tech. 50 15100 included in a special thesaurus Table 3: Possible increasing size of gencral lhesauri Although we realize that 50 or 100 terms are not enough in a special subdivision, say, librarianship, (see average number in a subdivision in Table 3), we cannot put in every term which we need because even a small in­ crease in each subdivision wil1 lead to a big increase of the overall size (see possible increasing number of descrip­ 1: The mostcornmonlyused terms, across all fields. tors in Table 3). Merely ten more terms for each subdivi­ 2: Vocabulary of each discipline or subject field, includes both terms that are likely to be used in onc field,and terms that sion will result in a term enlargement of thousands of will be uscd as common terms by other fields. terms. 3: Special vocabulary that will be used only in its special field,

Int. Classif. 17 (1990) No.2 -Zeng: Compatible General Vocabulary in China 93 (3) Conflict of updating and persisting. Since a general At any time, passive compatibility offers less capa­ thesaurus is often used as an indexing vocabulary or as a bility for establishing compatibilities among thesauri, reference vocabulary of many subject fields, the updat­ and vice versa. ing of it will have a wide influence. On the other hand, Despite a later start in thesauri development than since it covers vocabularies in several fields,the change of other developed countries, China benefitsa lot fr om ex­ any term becomes complicated because the relationships periences abroad, as well as from the standardization of among terms must be considered. This is why general the­ indexing languages in the world. The present fact is: sauri cannot so frequently be kept uptodate as special the­ - In addition to implementing international standards saun. and to develop national standards for thesauri, in 1979, the Chinese Documentation Standardization 2C: Conclusion of major obstacles in the compatible Committee recommended that all special thesauri approach between general and special thesauri should consider to establish compatibility with the Chinese Thesaurus. Most of the Chinese special the­ The major obstructions existing in the compatible ca­ sauri developed after 1980 used the Chinese The­ pability approach between general and special thesauri saurus as their basic reference source; some of them are all due to the traits associated with the compatibility also considered a compatible approach to other the­ of general thesauri. sauri in related fields. (I) Identification of equivalent concepts. As it was men­ Since the editorial group of the Chinese Thesaurus tioned above, a general thesaurus can neither reach as was composed of experts who became major editors high a speciality as a special one nor cover as much later for their special thesauri in their information special vocabulary as in a special one. As a result, it will centers, the same principles were adopted for special be very difficult,or, sometimes impossible to access equi­ thesauri. Almost all thesauri compiled by informa­ valent concepts in some special vocabularies. tion centers of ministries have similar principles in de­ sign. They are post-coordinated, discipline-oriented, (2) Concept environment structure design There exist quite a and for both manual and mechanized retrieval use. few scientific telms that represent different concepts or As a result, it can be stated that active compatibility is meanings when they are used in different subject fields. existing among Chinese thesauri. Therefore, itis amorecomplicated job to set up conceptenvi­ Let us take a further perspective on facts: By examining ronments for each term. Although broader terms(BT) and the broad classified display of two special thesauri and narrower terms (NT) may be identified according to the the Chinese Thesaurus, we get statistics about vocabu­ 'family' relationship, the related tenus (R T) cross references lary overlapping when considering major groups only. are hard to decide upon. No wonder: the second edition of Terms in nearly 80% of the broad classes and subdivi­ the National Defence Thesaurus of Science and Techool­ sion classes in the following two special thesauri are in­ ogy, which is a general thesaurus built on six special the­ cluded in the Chinese Thesaurus (see Column "B � A" sauri, gave up RT reference display in its fm al publication. and "B � A" in Table 5). (3) Coverage and sp ecificity balancing among different sub­ Another experiment was carried out on term conversion ject flelds. When establishing a general thesaurus by inte­ between the Aeronautics Thesaurus and the Chinese The­ grating several existing special thesauri (we will call them saurus by the information center of the Ministry of Aero­ 'source thesauri'5 ), a balancing between coverage and speci­ nautics. This experiment suggests that 572 (6.1 %) special ficity is hard to attain. Although we can design thecoverage, terms need to be added (see the 4th line in Table 6. Sinceac­ average specificity, and vocabulary size for each subject tive compatibility exists in both thesauri, more than 60% field, we can not design those for source thesauri. It is of the terms in the Aeronautics Thesaurus may be con­ possible that the existing thesauri will not adapt exactly to verted directly (see the 1st line in Table 6); and 482 (5.2%) our design framework. In this case, the choice of coverage special terms may be expressed by using post-coordinate and specificityfor each subjectfield, and the balance among terms of the Chinesc Thesaurus (see the 2nd line). dilTerentsubject fieldswill be very troublesome. Table 5 and 6 proved compatible capability between a general thesaurus and a special thesaurus where active 2D. Compatible capabilities among special thesauri compatibility exist. It is clear that although we will meet and general thesauri in China many problems caused by the internal contradictions of general thesauri as we try to construct a general Chinese vo­ The traits discussed above are certainly having their cabulary by integrating special thesauri, we might avoid impact on the degree of compatibility between thesauri. some barriers because of the existing active compatibility So far , there is no better answer available for us. But it among Chinese thesauri. does not mean no capability at all. Going back to Table 1, we believe that, if thesauri are designed with character­ 3. Awkward Compatible Capabilities Between Chinese istics which support the compatible approach, the final results would be different. There are two situations: Thesauri and English Thesauri (1) Thesauri are established separately and do not con­ 3.A Statement of the problems sider a compatible access (passive compatibility). As mentioned in the introduction of this paper, during (2) Thesauri are established following national and inter­ the years 1974-1989, approximately 40 thesauri devoted to national standards , by which thesaurus compatibility is special areas were developed, while more are being taken into consideration (active compatibility). planned and developed at present. However, this would

94 lnt. Classif. 17 (1990) No.2 -Zeng: Compatible General Vocabulary in China 1. A. Machinory Engin.. rlng Th�s�vrus B. CMnese Thesaurus duced to lexical, paradigmatic, and syntagmatic compati­

A < B* 6 , A A. - 6 Total bility; that is, compatibility in the representation of ob­ Bro�d Cl�" 12 4 3 19 jects, in hierarchical relations recognized, and in non-hier­ SuMlv13ion archical relations recognized, respectively. Structural com­ C18n 72 ( 61.02:'C ) 26 ( 22.03:'C ) 20 ( 16.90:'C ) 118(100:'C) patibility can be reduced to morphological compatibility 2. A. AeroMUU�s Thesaurus B Chmose' Th esourus (similarity in the structure ofterms) and syntactic compati­

A , 6 6 , A A - 6 Tolal bility (similarity with respect to the structure cf groups of terms) (5, p.186). These kinds of compatibility vary be­ Bro.dC!us 16 7 7 32

Subd!vlsion tween any two languages. Since the Chinese language has ••• 116 ( 54.8:'C ) 23 ( 10.8:'C ) 73 ( 34.4:'C ) 212(100%) CI an extremely different system both semantically and struc­ 4 Not., A

DIrect conversion available · 6,149 66.1% (2) Cultural background aspects. No matter whether a the­ saurus is designed with an orientation towards a collection Coordinated OT terms as equivalent 465 5.2% or towards a discipline, it reflects its social environment. Assigned to OT's broad terms 2097 22.5% Each word has its semantic meaning either in isolation, in Special terms need 10 be added 572 6.1% its context, orin its subject environment. In the fields ofthe Others 66 0.07% social sciences and liberal arts, a Chinese thesaurus might

Tolal number of descriptors in AT 9374 99.97% have a whole set of descriptors which are totally different from those in an English one. Problems in conversion may • Including terms that are equwalenl In both concepts and characters, and terms Ihal are come not only from a differentsubject coverage, but also equwalent In concepl5 bul different. In charac!9r5. from different political and ethical standards and nuances Table 6: Conversion experiment of AT and CT (source: (7» ofterms. In most fields ofsci ence and technology, there are less in­ only be one source ofthe general vocabulary that we are at­ fluences from the cultural background. However, in some tempting to build. It would not suffice because of a limited fields, there are still problems of subject coverage. For subject coverage and a limited emphasis on subjects in example, traditional Chinese medical science and medi­ these thesauri. cine are very important in China, but terms in these fields Thesauri in other languages could be considered as an­ are excluded inmost English medicine and related thesauri. other source of a Chinese general voc'abulary where the­ Cultural background aspects cannot be avoided and sauri published in English could be considered as the will influence the compatible approach. We might find major source because most indexing and abstracting ser­ rules to solve problems raised by multi-lingual aspects, but vices, especially those in machine-readable form, are pro­ it seems to be very unlikely to find rules to solve problems duced by English-speaking countries (5, p.217). raised by the cultural background, especially when ma­ chine automated mapping 01'conversions are considered. There exist quite a few bilingual and multilingual con­ trolled vocabularies in the world. As early as 1971, a biling­ 3_C Internal features of Chinese thesauri ual (English/Spanish) index to six such vocabularies in the field of biomedicine was developed. MeSH now exists in and their impact on compatibility several languages, allowing MEDLINE to be interrogated (1) Non-romanized characters as main entry. Here is a typi­ in languages other than English (5, p.229). Chinese the­ cal entry in a Chinese thesaurus: sauri often have a part called an English-Chinese bilingual

index, but these indexes are based on a translation of Deep struclUre Chinese descriptors. As a result, the English terms and Use Undertylng structure phrases are by no means equivalent to English descriptors. Underlying structure There is no example of conversion or mapping between a UF Deep structure Chinese thesaurus and an English one; and no research has NT left·branchlng construction aT Transformallonat·generaUve been done on this topic. grammar e · IT Grammar Wh ther it is possible to achieve compatibility between RT Surface structure existing Chinese and English thesauri is still a question. To examine their compatible capability, aspects and features To display descriptors in the main part of the thesaurus associatedwith this question will be discussed in the follow­ (The alphabeticel list in an English thesaurus) and in the ing parts. hierarchical subject category indexes, there are various ways: 3.B Aspects which determine compatibility a. Arranging terms In an order of slrokes observed In calligraphy, say, " ;� between Chinese and English thesanri (deep) has 11 strokes, while "ffi:. (under) has eight strokes and should be listed before (1) Multi-lingual aspects. Glushkov et al divide forms of compatibility into two, i.e. semantic compatibility and structural compatibility. Semantic compatibility can be re-

Int. ClassiC 17 (1990) No.2-Zeng: Compatible General Vocabulary in:China 95 b, Arranging tarms by their radicals wUh strokes, observed in calligraphy, tion, The maximum length of a descriptor is limited to say, 'i'Jf..'(daep) belongs to ' 5- • radical wllh nIna more stokes, while .Iff....(under) around 15 Chinese characters, Even for a human being

belongs to • t, radical wllh fiva more strokes. Radicals ara usually arranged by the (instead of a machine), it is hard to get the consistent re­

dlrec\lon of the first stroke, sult for keyword separation in such descriptors. (4) Compound terms with a strong cultural background. It Both of these arrangements have been common in is not strange that any thesaurus in any country will re­ Chinese dictionaries and bibliographies for hundreds of flect its cultural background, butin Chinese thesauri, pol­ years, but these two methods require users to examine itical and cultural influences seem much stronger than in each word every time to calculate the number of strokes thesauri in other languages. In addition a Marxism-Le­ of radicals and are not adopted to computer use. ninism bias and slant are strongly encouraged. This fac­ c. Arranging terms by romanl�ed leiters. Each word Is presented by tor might be one of the biggest obstacles to the com­

romant�ed leiters aCcx.'rdlng to lis pronunclallon In standard mandarin Chinese, say, ';';f. • patible approach.

• Is presented as 'sMn', while 'fi\' Is presented as "dl' and should be listed before

"shen". 3.n Compatibility between Chinese thesauri There are three romanization systems for Chinese charac­ and English thesauri ters of which the Pinyin system is most widely accepted in With the exception of terms with a strong cultural back­ the world. However, it cannot replace other methods be­ ground, we are glad to have found that compatible capa­ cause people aged over 40 are not familiar with it; and bilities between thesauri in these two languages exist be­ most people in the South, East, and West of the country cause of the thesaurus structure and the academic termi­ have problems in pronouncing Mandarin correctly. nologyaspects. There are other ways to arrange Chinese characters. (1) Unified thesaurus structure. In the second part of this Presently romanization is recognized as the best way for paper, we explained the concept of "active compati­ information communication and is used in thesauri and bility". Chinese thesauri have benefited a lot from inter­ bibliographies. Some thesauri attach indexes in which national standards for documentation which include characters are arranged in different ways. those for thesaurus construction. They fit into a structu­ ral framework similar with that of other countries. Flexi­ (2) Multiple choices fo r alphabetical display. Even when bility exists in the degree of term precoordination and romanized letters are chosen to be the method of arrange­ term fo rmation but less so with respect to sbucture. A ment, there are different alphabetical display principles. structure which is composed of an alphabetical list as the Letter-by-Ietter is used by the Chinese Thesaurus be­ main part, a subject group display, a hierarchical display, cause ofthe complicated word parsing process. Word-by­ and a bilingual display as indexes, as well as a set of sym­ word is used by many special thesauri because they want bols for a cross-reference system, are commonly ac­ to get terms that have the same beginning characters cepted for the construction of thesauri. Such a structure together. Word-by-word is welcomed by both users and was only developed after the Chinese thesaurus had been professionals, but it needs much more work on thesauri established. compilation than does letter-by-Ietter. Other kinds of al­ phabetical display are used by some thesauri, and each (2) Academic terminology agreement. According to re­ has its own advantages and disadvantages. search work done by Chinese linguists, Chinese aca­ demic tenninology has three characteristics which are (3) Difficulties fo r machine recognition. In Chinese helpful in international communication. words, sentences, titles, and texts, all characters are put one after another except when a punctuation mark ap­ (2a) Chinese possesses monomorphemic words, mainly pears. Ten years ago, when KWIC and KWOC were con­ the names of chemical elements, organic chemical com­ sidered for Chinese indexing and abstracting agencies, ex­ pounds, units of measurement, and words translated perts found many difficulties related to the structure of from foreign languages. Such words usually have Eng­ the Chinese language. It is hard for a machine to recog­ lish equivalencies. They are easy to translate. nize keywords in a title in which no sign or space exists be­ (2b) Chinese has compound words which form the ma­ tween any two characters. This is also true in term and jority of academic terms. There are six kinds of com­ phrase recognition. A compound word might be separ­ pound formation in the Chinese language. Most of the ated in several ways to give different translations. academic compound words fall into the following two

e-a a, Shengtushuguan provIncIal library kinds offormations: 4iIfjtir. b. Slieng Tuslluguan province + library Part I modify part II (I -- II) /Ii ifJ"'" 1. > . Here there are two situations:

c. Shsng Tushu Guan Province + Book +Bulldlng BC d. Sheng Tu Shu Guan Province + Picture + Book .g. -'> • = . + House • � 1- .J� f� .f.·,);;-' 1 gl �. Il; infInitesimal �.> calculus Since thesauri are designed for both manual and mech­ anized retrieval systems, more than 50% ofthe entries in AB --> C , •.g. tube

96 Int. Classif. 17 (1990) NO,2 - Zeng: Compatible General Vocabulary in China 2. Words with prefIx or sufflx 4.2 Design of concept (includes its expression in a the­ saurus) environment structure. Once an equivalent con­ e.g. .antlfreeze �t+fr.J cept is attained, it is necessary to design an environment .JX'!,\ 'rt. modernization structure which consists ofits hierarchical and other non­ hierarchical relations. It is hard to attain compatible and These two kinds of words have a less vague meaning and proper results especially when morphological and syntac­ are easier to translate than words belonging to the other tic compatibility is also considered at the same time. four kinds offormations. 4.3 A balance between coverage and specificity. Since each of the source thesauri has its emphasis and orienta­ (2c) 90% of the phrases in Chinese academic termino­ tion, it is hard to attain a balance between coverage and logy fit into the formation " I --- II"(i.e. Part 1 modifies specificity. part II), most of them are "fixed" phrases, that means, 4.4 Reflection of internal fe atures of the Chinese lan­ their forms are set by a terminological authority and do guage on Chinese thesauri. There are problems remain­ not vary with their use. For example, ing in Chinese thesauri construction, such as entries and their arrangement, and their recognition by machines. " li.:J1 "is used for "Index" instead of R �j f1 "and " .{.j,.I;/ ". On the other hand, strong cultural and social back­ It is possible to find rules for academic terminology grounds have such important influence on compound conversion and mapping. This might be the point from terms that it is unlikely to attain as many compatible re­ which we may start. sults among terms in social science and liberal arts as in scientific and technical fields. (3) Similar relationship between a top term with its nar­ However, although there will remain at least as many rower tcrms. It is very interesting to measure the relation­ problems as we have discussed in this paper, there exists a ship between a top level term and its narrower terms. compatible capability among Chinese special thesauri Table 7 shows the percentage of trees among some the­ and general ones, as wen as between Chinese thesauri sauri in which all narrower terms have a mutual root and English ones. From the analysis above, we found word which is either the top term or part ofthe top term. that most ofthe Chinese thesauri were established fo llow­ ing the national and international standards, by conside­ Table 7. Trees in which top term and narrower tenns have mu­ tual word roots. ring compatibility with the Chinese Thesaurus and some English thesauri in related subject fields. Therefore, it Title Number % seems to be possible that some ofthe problems discussed INSPEC Thesaurus 234 (English) 48 above may be avoided, or, may - at least - be reduced to a ' Chinese Thesaurus 483 (Chinese) 54 certain extent in the process of establishing a compatible vocabulary. Aeronautic Thesaurus 82 (Chinese) 49 No doubt order to establish such a compatible vocabu­ Machinery Engineering Thesaurus 442(Chinese) 70 lary, we need fu rther research. First, we need to know National Defence how to apply terminology. semantics, as well as other lin­ Thesaurus ofSci. & Tech. 144 (English) 26 guistic theories and methods for the process of establish­ ing such a vocabulary. Second, we need a survey and a Once a root has been translated correctly, whole comparative study of the commonly used thesauri pub­ words in the tree may share the result. This is an advan­ lished in China and abroad. especiaUy in English-speak­ tage not only for translation, but also for machine recog­ ing countries. Third, we need fu rther perspectives for the nition. as well as for the application of KWIC and feasibility of establishing a compatible general vocabu­ KWOC to Chinese indexes. lary in China. With such an effort, we might convince the related agency and our profession to realizc the import­ 4. Conclusion ance ofthis project. From the discussion above, we can begin to recognize After summarizing the status of compatible capa­ the problems and capabilities existing in the compatible bilities, I would strongly suggest to begin a project for a approach for the establishment of a compatible general compatible general vocabulary in China. While a lot of vocabulary in China. Since such a vocabulary is sup­ outstanding work has already been done by researchers posed to be based on existing thesauri (most of them are in the field of indexing languages in many countries, the special thesauri) published in China and in English­ establishment of compatibility of the Chinese indexing spoken countries, it will be confronted with obstacles due languages must be done by Chinese and a cooperation to aspects associated with general thesauri and multi-lin­ with researchers in the world will also contribute to the gual thesauri. Conc1udingly we will list some of the major field and improve the attempts at resource sharing obstacles: among human beings in the whole world. 4.1 Identificationof equivalent concepts either between a special thesaurus and a general one, or between a Noles Chinese thesaurus and an English one. In this process, size, specificity, coverage, as well as cultural and social In the report of Unesco: "UNISIST Study Reporton the fea­ sibility ofa world science infonnation system", 'Compati­ background of the source and target thesauri will have a bility' is definedas: A quality ofsystems whose products can great influence on the compatible approach. be used interchangeably, notwitchstanding differences in no-

Int. Classif. 17 (1990) No.2 - Zeng: Compatible General Vocabulary in China 97 tation, structure, physical carriers, ect., without any special (4) Kratochvil, Paul. The Chinese language today : Features of "conversion machinery". an emerging standard. London: Hutchinson Univ. Library 'Conversion' is definedas "The process of transfonning in� 1968. formation records, with regard to transcription encoding, (5) Lancaster, F. W.: Vocabulary control forinformation retrie­ data structure, etc., so as to make them interchangeable be� val. 2nd ed. Arlington, VA: Infonnation Resources Press tween two or more services or systems using different con� 1986. ventions and media".(Glossary,p.l47). Source: Dahlberg, (6) Levy, F.: Compatibility between classifications and the� 1981 sauri: Evaluation of a first study in the field of information 2 See Lancaster, 1986, p.184. storage and retrieval. Paperprescnted at the 3 See Lancaster, 1986, p.l87-188. Conf. Int.Federat.Doc., Tokyo 1967. 4 See Lancaster, 1986, p. 187-188. (7) Qiu, Feng: The relationship between gcncral thesauri and 5 The tenn and the idea was introduced by Dagobert Soergel special thesauri. Infonn.Science, No.2(1982) (In Chinese) in his book "Indexing langyuages and thesauri: construction (8) Soergel, D.: Indexing languages and thesauri: construction & maintenance" in 1974. & maintenance. Los Angeles: Melville 1974. (9) Wall, E., Barnes, J.M.: Intcrsystem compatibility and con­ vertibility of subject vocabularies. Philadelphia, PA: Auer­ bach 1969. (Technical Report 1582-100-TR-5) References (10)Zeng, Lei: Establishing a unified system of descriptor lan­ guage for use in China. J. of the China Socicty for Sci. & (1) Dahlberg,I.: Conceptual compatibility of ordering systems. Tech. Information 6(1987)No.l, (in Chinese) Int.C1assif. 1O(1983)No.11,p.5-8 (11 )Zeng, Lei: An introduction to thesauri and classificationsys­ (2) Dahlberg, I.: Toward establishment of compatibility be­ tems in the People's Republic of China. Int. Classif. tween indexing languages. Int. Classif. 8(1981)No.2, 13(1986)No.1, p.24-28 p.86-910 (3) Hammond, W., Rosenborg, S.: Experimental study of con­ * This research was undertaken under the instruction of Dr .Edie vertibility between large technical indexing vocabularies. Sil­ Rasmussen. The author would like to thank Dr. Rasmussen for ver Spring, MD: Datatrol Corp. 1962. her great help in this research and the editing ofthe paper.

Information, Data, Knowledge, Classification lagen. - P.O.DEGENS, Dusseldorf: Implizite formale and Structuring Modellvorstellungen in den Methoden der phylogeneti­ schen Systematik. - W.H.E.DAY, St.John's, Canada: These topical concepts were made the theme of the Estimating phylogenies with invariant functions of data. 14th Annual Conference of the Gesellschaft fUrKlassifi­ - E.A.SCHACHTEL, Stanford, CA: EinfUhrung in das kation eV held in Marburg, FRG, March 12-14, 1990. Human-Genome Project. H.J.BANDELT, Prof. Peter IHM and his coorganizers, Prof.H.H.BOCK A.v.Haeseler, J.BOLIK, H.SCHUTTE, Maastricht et and Dr.H.-J.HERMES have done anenormousjob to as­ al.: A comparative study of rRNA homology data: se­ semble a rather huge group of people (some 95 papers) quence similarities versus evolutionary distances and 5S who presented their papers on current research in most rRN,A versus small subunit rRNA. different fields. There were three plenary sessions, the first one In addition there were sections and workshops with together with the Society for Multivariate Analysis in the the fo llowing topics: Classification and Structuring in Behavioral Sciences (who happened to meet at the same Art and Archaeology; Numerical Classification; Con­ institution on the very same day) with the papers by cept Representation; Classification in Administration H.GLASHOFF, Hannover, on "Analyse von Schadens­ and in Economics; Graphical Representation of Simi­ symptomdaten mittelalterlicher Wandmalereien", and larity Structures; Classificationand Data Analysis; Con­ G.ARMINGER, Wuppertal, on "Datenanalyse mit cept Analysis and Structuring of Symbolic Information; dem linearen Modell: M6glichkeiten und Grenzen unter Databases in Museums and in Historic Research; Data Beriicksichtigung neuerer Entwicklungen". - The sec­ Analysis in Archaeology; Terminology and Indexing; ond plenary meeting was devoted to Tenninology and Classification and Data Analysis in Medicine and Biol­ Databases with papers by W.NEDOBITY, Vienna, on ogy; Libraries and Archives; Analysis of Spatially Or­ "Die Rolle der Klassifikation in der Tenninologiedo­ ganized Data; Expert Systems and Computer Supported kumentation" and Ch.WOLTERS, Berlin, on "Objekt­ Procedures; Classification and Knowledge Handling in datenbanken uud Thesauri fiir kleine Museen". - The Medicine; Knowledge Organistion in Databases; Biome­ third plenary meeting on "Classification, Systematics tdcal Problems in Genome Sequencing; Numerical and Evolution" was organized together with the German Classification and Biological Taxonomy; and Ordering Chapter of the International Biometric Society and had and ClassificationProcedures in Medicine. five papers, almost all on problems of phylogeny: For further information please contact: Prof.Dr.Peter O.KRAUS, Hamburg: Phylogenetische Systematik. Ein­ Ihm, Institut fUr Medizinische Biometric, Bunsenstr. 3, fii hrung und Darstellung ihrer methodischen Grund- D-3550 Marburg.

98 lnt. Classif. 17 (1990) No.2 - Zcng: Compatible General Vocabulary in China tiveness. Further research will examine table-of-contents infor­ mation in other subject areas, as well as other sources ofcontents infonnation such as abstracts and back-of-the-book indexes.

Reports Nicholas BELKIN; Tefko SARACEVIC, Rutgers Univer­ sity, School of Communication, Information, and Library and Communications Studies: Design principles for tbird-generation Online Public Ac­ cess Catalogs taking account of users and library use. This project will establish design principles for Online PublicAc­ cess Catalogs (OPACs) on the basis ofempirical study of library users and library us'e. Our maj or assumption is that OPACs arc tools to support people as they see information. We must under­ Pertinent Research at OCLC stand how people intend to use information, how they seek it, In January 1990 we received the "Annual Review of and how users' characteristics relate to these issues. Based on this data, we will suggest functions that OPACs should and OCLC Research, July 1988-June 1989", 68 pages fu ll of could perfonn in various circumstances, identify the resources most interesting information. In his introduction Martin and the information about users such systems would need, and DILLON, Director of the Officeof Research of the On­ propose some principles for overall OPAC design. line Computer Library Center, Inc. in Dublin, Ohio Karen Markey DRABENSTOTT, University of Michigan, stated that the Office focusses at present on four strategic School of Information and Library Studies; Diane VIZINE­ areas: (1) enhancing use ofthe OCLC Online Union Cata­ GOETZ. OCLC: Increasing the accessibility of the Library of Congress Subject Headings inonline bibliograpbic systems. log; (2) investigating the requirements of the library of OCLC and the University of Michigan engaged in a collabora­ the fu ture, particularly in an academic setting: (3) increas­ tive research project to investigate automated techniquesto link ing the productivity of catalogers; (4) digitizing data, subject terms entered by online catalog users with the library's controlled vocabulary. This report describes comparisons be­ both to preserve library materials and to transfer infor­ tween assigned subject headings from a large bibliographic file mation fr om paper fann into automated retrieval sys­ and LCSH-mr (machine-readable Library of Congress Subject tems. - At present the OCLC Online Union Catalog has a Headings) and comparisons of patron entered subject access points and the library's controlled vocabulary, phases 2 and 3 of database of more than 20 million bibliographic records. the project, respectively. The final project report will include rec­ Many of the projects performed in the Officeof Research ommendations for subject searching improvements to guide on­ aim at enhancing the use of this database. But there are line catalog users from their subject access points that are not ex­ pressed in terms of the catalog's controlled vocabulary to also projects investigating the "Library of the Future" LCSH-mr and its syndetic structure. It will also provide display and the possible increase of the productivity of cata­ guidelines for LSCH-mr, its syndetic structure and linked as­ logers. In the following the nine (out of25) pertinent con­ signed subject headings. tributions are listed with their project investigators and Clement YD, University of Illinois at Chicago, Department the abstracts to the project reports given in the Annual of Electrical Engineering and Computer Science: Learning term Review. Of interest is also the Distinguished Seminar weights. Learning and clustering procedures to weight terms for retrieval Series. Amongthe 6 seminars the one given by Dagobert are described. Experimental results using two document sets SOERGEL should be mentioned. It tells about his re­ show that a method oflearningtennweights based on the nonbi­ search project "Improved Access to LC Classification nary independence model gives better performance than the in­ verse document frequency method and that learning from re­ and LC Subject Headings for Online Searchers and Cata­ lated queries, instead of from arbitrary queries yields improved logers". The responsible Research Advisory Committee retrieval performance. A formula to incorporate the frequency members oughtto be congratulated for having provided oftenns in queries is presented. valuable input indeed in setting the course for the OCLC Nancy J.WILLIAMSON, University of Toronto, Faculty of research activities during this period. Library and Information Science: The Library of Congress Classification in the Computer Age. Research leading to a detailed content analysis and profileof the Stuart L.WEIBEL, Research Scientist: Applying neural net­ Library of Congress Classification schedules is presented, in­ works to c1assification problems. cluding a description of the design, methodology, and prelimi­ Connectionist systems, often referred to as neural networks, repre­ nary findings ofresearch in progress. Final results ofthe study sent a promising method for patternre cognition and classification. are expected to aid the Library of Congress in the development Thisreportdescribes the application ofneural networks to theprob­ of a MARC (MAchine Readable Cataloging) format for its lernof automated classification of document layout objectsas text classification schedules and in the conversion of the schedules orextraneousnoise. Thesemethods arecomparedwiththeperfonn­ into machine-readable fonn. ance of a conventional statistical method of classification and the advantages and disadvantages ofeach are discussed. Gerard SALTON, Cornell University, Department of Com­ puter Science: A syntactic approach to automatic book indexing. Roland HJERPPE, Visiting Scholar: Bibliographic control Automatic publishing systems are now widely used to produce and document architecture in hypermedia databases. books and documents of many types, and the text of these ma­ This project focusses on the need and potential (1) for designing terials, available in machine-readable fo rm, can be used for auto­ components and structures (fonnats) for electronic documents matic processing. This study describes automatic methods, to that enable algorithmic derivation and generation ofa variety of generate back-of-the-book indexes based on a syntactic analysis bibliographic representations that can be used in "catalogs" of of the text and a phrase-generation system that identifies collections of electronic documents, and (2) for designing these meaningful index entries. catalogs as adaptable instruments integrated with the other tools available to the users or producers ofelectronic documents. Roy R.LARSON, University of California, Berkeley, School of Library and Information Studies: Enhancing topical search­ Martin DILLON, Director, OCLC Officeof Research: ing using classification clustering. Enhanced bibliographic retrieval. Early investigations of user interaction with online library cata­ This project evaluates the retrieval effectiveness of contents in­ logs,showed a dramatic increase in subject searching when com­ formation in bibliographic records. The first stage ofthis project pared with card catalog use studies. However, subject searching examined table-of-contents information in a database of pre­ also caused the highest rates of search fa ilure and presented dominately technical monographs. The results showed that the in­ many difficulties for users. A method of enhancing topical sub­ clusion ofthis infonnation significantly improved retrieval effec- ject searching using classificationclustering is discussed.

Int. Classif. 17 (1990) No.2 � Reports and Communications 99 Preliminary Programme TKE'90 sisted human translation.·-ASCHILLER, P.STEFFENS: A two-level envi­ ronment for morphological descriptions and its application to problems of The Second International Congress on Terminology German inflectional morphology. - Y.UEDA, S.AHLEN, K.KOGURE: Declarative control of generation using typed fe ature structures. - and Knowledge Engineering, 2-4 Ocl.l990 in Trier, Ger­ T.OSADA, S.MATSUYAMA, M.NAKADATE, H.NAKAI, T.1SHIBE: many promises a lot. Again, as already in the first con­ An analysis of drug intneractions using descriptions in drug package insert ference of 1987 a big number of Organisations and In­ documents; a study on standardizing therapeutic words.RL.BUCHAN: Retrospective indexing (RI) - A computer-aided indexing technique. - stitutions support this conference which will - if its pre­ M.T.Genuardi: Knowledge-based machine indexing from natural language liminary program holds what it seems to promise - text knowledge base design, development and maintenance. -N.J .I.MARS, become a major event! There will be the following nine H.Yan der VET: Using abstracts for semi-automatic knowledge acquisi­ tion. -S,SIMONIAN: Conceptual description ufthe linguistic information keynote lectures: A.DZINCHARADZE: The role ofter­ base of modern Armenian. minology in creating knowledge bases. - E.OESER, Sessio1l 4: Documentalion Languages alld Ordering of Knowledge. Chair: W.SCHMITZ-ESSER, B.LUTES-SCHAAB: An online thesaurus for fact University of Wien: Terminology science and knowledge retrieval. -T.ROGOZHNI KOY A: Individual infonnation thesaurus: asso­ theory as a prerequisite of knowledge. - A.M. TJOA, ciative fields and a model of semantic development. - F.SARRE, J.MIT­ R.WAGNER: Basic conceptual elements of knowledge TERMEIER, U.GDNTZER, G.IDTTNER: Evaluation of the thesaurus generating information retrieval system TEGEN with regard to learning be­ based systems. - W.RAUCH: Information science in an haviour and user acceptance. - F.KOYACS: Terminology and conceptual information society. - A.MELBY: Terminology and system for creating patient records. -RSUNEETI: Thesaurus in law: some technical documents' the importance amI limitation of problems ofconstruction. -S.SEETHARAMA: Towards establishing con­ cordanceamongmedical classificationsystems.-P.STANCIKOY A: Biling­ formal systems. - M.SCHAAR: Multilingual informa­ ual thesaurus YODOINFORM and its new fu nctions in application of tion and knowledge management. - D.WALKER: Lexi­ CDC/lSI software (micro- and mainframe). - R.YOLHO-LOPES: Auto­ cal resources for the management of massive text-files. - mated access to multilingual information. Session 5: Electronic Dictionaries. Chair: T.YOKOI. G.ENGEL: New professional profiles in knowledge en­ L.BELIAEYA, RPIOTROWSKI, S.SOKOLOYA: Principles of linguistic gineeringand knowledge transfer. automata and their information bases design. -J.-M.MAES: The intelligent dictionary project. - RMAYER, D.GEER, K.-H.HANNE: On how to Furthermore, 8 sessions are foreseen with altogether bring Hypertext to termbanks. - S.MIIKE, SAMANO: The structure and 85 papers. 7 workshops will be held as preconference fu nction oflhe EDR concept dictionary. -A. YEZHO Y: Organization oflex­ events, most ofthem have'already' been mentioned in the icographic data on the basis ofa lexical unit oriented approach. -P.BOUIL­ LON, L.TOYENA: Word fo rmation and computational dictionaries. -Kim note on the 2nd Call for Papers, TKE'90, published in DEOK-BONG, Hyun-Jae 1M, K.CHIO, G.KIM: KOCP: an extended OCP Int.Classif.89-3, p.162. The Session papers will be listed for Korean and English.. .. L.LEMNITZER: A conceptual schema for a multifunctional lexical database: a proposal. - J.GMERLO, deS.ZITZER: in the following: Terminological electronic dictionaries as tools for aceess to knowledge .. Session 1: Te rmillolog)1 /Vww/edge Theory alld Knowledge Ellgineering AA.POLIKARPOY, L.KOLODYAZHNAYA: A system for compilation New Applications. Chair: E.OESER and analysis of electronic philological dictionaries using a personal com­ O.LASKE: Acquisition of knowledge for expert systems: unsolved prob­ puter. -J .RAY, M.NADKARNI: Electronic dictionaries- an Indian experi­ lems (working title) . .,-N.P.PESHKOVA: Different types of discourse struc­ ence. -Y . TANAKA, S.YOSHIDA: Preparation of a concept dictionary. ture as a means of knowledge representation. -J .HAAS: Treatment of uncer­ Session 6: biformptioll Managemellt ill Organizatiolls. Chair: H.CZAP tainty in social-science expert systems. - G.BUDIN: Scientific knowledge H.CZAP: Construction and representation of concpets in enterprises. structures. - A.ASTOCHIK, S.NIZHNY, S.: Some clear and unclear as­ M.RSHACKLETT: Integrating information systems into corporate pects of research of relationship. - K.AYBRBUKH: Term as a subject of strategic planning and plan execution. - H.BERTRAND-GASTALDY, study and a tool for knowledge fixation and knowledge transfer. - L.C.PAQUIN, L.DUPUY: The need for information and knowledge man­ B.SORENSEN: 'Level Structuring' of knowledge and its application to an agement. -D.DAOUST: The evaluation ofsoeiolinguisticand terminologi­ expert system domain with a particular view to the perspectives for termino­ cal change in a commercial enterprise. logy work in general. - S.D.SELOY: Terms, tennability and knowledge. ­ Sessioll 7: Compllter Support ill Te chnical Commllnication. Chair: F.ACITKINA: Comparative terminology theory: problems" goals, S,-E. WRIGHT methods, applications. - B.Y.GORODETSKY: Cognitive aspects oftermi­ H.EISELE: Les fr ustrations de tcrminologue. -UHEID, G.FREIBOTH: nological phenomena. - S.V.GRlNEV: Terminology and knowledge the­ Terminological and lexical knowledge for computer-aided translation and ory. - M.F.PESCHL: Some cricial reflections on symbolic knowledge rep­ technical writing. - R.G.REIN: Hypertext: indispensable information sys­ resentation. - T.PETROVA: Gnosstic aspects of the term. - K.URAZ· tems for communicators. -H.-D.LUCKHARDT: Automatic terminology BAEY: Space (exploration) terminology and the integration of science and support in computer-aided and machine translation. -G.M.SHREYE: Re­ engienering sublanguages.B.I.GUREVITCH: The descriptive approach to quirements analysis, empirical research and prototyping in the software en­ visual knowledge representation and visual processing. gineering of workstations for computer assisted translation. - Session 2: Knowledge-based Systems. Chair: W.KLAS S.-RWRIGHT, C.LARSON: Building the universal engine: implementa� K.BAKARSIC: Science citation index for the field ofarcheology hermeneu­ lion oftranspaltfonn MTX. - S.RDIAMESSIS: Ongoing efforts in Greece. tical and informatics aspects of SCI based systems of regional bases. - -S,SIRAYEYA: Compound term engineering. - L.L.NELJUBIN: Termi­ W.BENN: Modelling mulliple paradigm support in a KBMS. - R.FU­ nographic text formalization. JIWARA: Intelligent design of control systems' combining knowledge­ Sessioll S: lermillology alld lVlOwledge Transfer To ols. Chair: C.GALIN­ based systems with conventional design methods. - M.LINSTER, SKI B.GAINES: Supporting acquisition and performance in a hypermedia envi­ W.DILGER: Transferring knowledge bases among hybrid knowledge rep­ ronment. -I.MEYER, D.SKUCE: Computer-assisted conceptual analysis: resentation systems. - AKUKULSKA-HULME: Speed understanding of an essential component of a terminologist's workstation. - J.NIGGE­ an unfamiliar domain. - M.OSADTSCHUK: Utilization of semantic MANN: Analysis and representation of neuroanatomical knowledge. - frames instudying and teaching terminology. -M.1DIEGUEZ,A.M.BUR­ G.OTMAN: Semantic networks in terminology: the case ofAl.- M.A.Par­ DACH, P.l.CABRERA, RC.G.GARBARINI, V.P.HORMANN, tyka:The application of the Quine-McCluskey algorithmm of self-assigning R.LAZO: A terminological microbank for Chile and Latin America. - minimization to knowledge-based systems. -J.WALLMANNSBERGER: C.GALINSKI: Knowledge transfer to/from countries with languages using Hypertext approaches to terminological information processing. non-European scripts. - F.GAUDIN: La socioterminologie et les discours Session 3: Natural Language Processing and Knowledge Engineer­ d'experts. - L.GUESPIN: Socioterminology fa cing problems in standardi­ ing.Chair: B.B.RIEGER zation., - CH.HOSBAYAR, HASBAGAN: A study on Mongolian termi­ B.ENDRES-NIGGEMEYER: A cognitive model of abstracting. -Ch.DE­ nology for physical concepts. -D.M.ZIYADU LLAEYA: Thematicdiction­ FRISE/S.NIRENBURG: A language for representing text meaning. _. ary on hydrotecnical construction. - W.NEDOBITY: Methodological and T.BULOT: Generateur de dialogue et optimisation de la consultation tcU:­ practical aspects of computer-assisted knowledge transfer to developing matique. - G.KLOSE, K.LUCK: The representaiton of knowledge in countries. -J .SANS: EOQL: Glossary of terms used in the management of ULOG. - M.TARNOWSKI: Text-Understanding- extracting knowledge quality. -Wubin SU: China National Technical Committee on Termino­ from written texts. -M.GERLACH: Terminological reasoning in the natu­ logy for Standardization (CNTCTS). ral language consultation system WISBER. -J.-U .MOLLER: How to com­ Therc will bean accompanying exhibition. The proceedings arc to be avail­ municate expert knowledge. -H.SORENSEN: The use of knowledge-based able to participants at the beginning of the Congress. frames for terms in EUROTRA. -I.SEREBRY AKOYA: Non-neural lexi­ For further information contact The TKE'9 1 Congress Secretariat, cal units in scientific texts and the problem of their human and machine-as- Dipl.Kfm.Thomas AOchs, P.O.Box 3825, D-5500 Trier.

100 Int. Classif. 17 (1990) No.2 - Reports and Communications First Meeting ofthe ISKO/GC (German Chaper) WG In­ dexing, Classifying, and Retrieval ISKO News 3 On Monday, Aug.13, 1990 this WG will meet in Dann­ stadt. Its chairman had proposed to discuss the problem, "why in comparative studies intellectual indexing often ­ apparently or as said or actually - has not proven better First International ISKO Conference, than automatic indexing". For further information on Darmstadt, Aug.I4-17,1990 this meeting please contact: Dr. Robert Fugmann, Alte In addition to the papers of the conference program Poststr. 13, D-627 Idstein. listed in the previous issue we are happy to announce that Mrs.Pat MOLHOLT, Rensselaer Polytechnical In­ SCCAC News stitute. Troy, NY will also participate with a paper, dis­ The Societey for Conceptual and Contents Analysis cussing the findings on the related term structure for the by Computer" has just issued its Sth Newsletter, starting Art and Architecture Thesaurus. She had identified over out with the good news that the Society will continue to 100 relationships and will speak about them in" Standar­ operate "for a while" on the same bases since the founda­ dizing and Codifying Related Te rm Linksfor Improved In­ tion ofthe Society in 1983, that is, without chargingmem­ fo rmation Retrieval". bership fees thanks to the offer from ZUMA e.Y., Mann­ The preparations for the "International Soft ware Ex­ heim to "help organize things fr om there". hibition for Th esaurus, Classification, 'Terminology, and In characterizing its aims it is said: Indexing Applications, 14-16 August 1990, TH Darm­ SCCAC is a loosely organized group of scholars in the hu­ stadt" have started. Mr.Horst KORNER, ABD Info manities and related disciplines sllch as the social sciences and psychology as well as communication sciences interested in the Consult, Zittelstr.S, 8000 Munchen 40, has taken over its sp ecial area of computerized- colllelll and conceptual analysis. organization. Please help to infonn everybody whom This area is very generally definedas including method� of lexico­ you know ofto have produced such software that this in­ graphy, indexing, and text analysis US;'lg entire conceptual sys­ tems such as Roget's Th esaurus rather than alphabetical systems formation will reach him/her. of ordel: It includes automatic retrieval systems and thesaurus building as lVell as studies searching fo r motifs alld themes in In the series of visits to institutions on Friday, Aug.I7 larger text corpora. It also includes content analysis using in the afternoon which were announced in the program Jp ecial dictionaries for text analysis, which fo llows models like and described in detail in a list sent to the registered par­ the GENERAL INQUIRER and Iker's WORDS. It shares with Artificial Intelligence and Machine Translation the interest in ticipants a replacement for the ESOC visit has been classification and representation of natural language meaning made. An excursion is foreseen instead as an invitation and the building oj knowledge bases. by KTS Information Systems, Munich to a free busride Thefuflction oj SCCAC is to provide a mediumjor communica­ to the German-French Institute at Ludwigsburg bei tion and injormation to scholars working in the various disci­ plines within the given commOIl areas of interest with the least Stuttgart for a demonstration of a Multilingual The­ amount of beaurocratic red tape and financial burden. SCCAC saurus in the field of 'International Relations and Coun­ will provide addresses of participating scholars, disseminate pro­ tries', integrated into databases with dynamic develop­ ject descriptions, plans, and injormation via newsletter, and or­ ganize meetings on a regular basis, sometimes ill conjullction ment possibilities, using the software DOMESTIC. with ACH (Association jor Computers ill the Humanities), Please note that the ISKO General Assembly will take ALLC (Association fo r Literary and Linguistic Computillg) Conjerences and ISKO (International SOciety fo r Kn owledge Or­ place in Darmstadt onAug.IS, 1990, 6-7 PM. The invita­ ganization). SCCAC is also a recognized sp ecial interest group tion for this together with the Agenda is part of ISKO of ACH. ALLC, ACH, and/or ISKO memberships are recom­ News 2 in Int.Classif.90-1. We are looking forward to see mended but not requiredfor SCCAC members. you at this assembly and hear your opinion since at this occasion the discussion offuture work ofISKO is among The Newsletter contains the abstracts of contributions the Agenda points. to the 2 volumes edited by Klaus SCHMIDT: Content - At some date during the conference -still to be an­ Concepts - Meaning. Publications from the Society of Conceptual and Content Analysis by Computer nounced - the members of the Gennan ISKO Chapter (SCCAC). Greenwich, CT: JAI Press 1990. (Advances in will assemble to plan the next (the second) national con­ Computing and the Humanities, vols.3 + 4), which give ference in 1991. It is understood that this First Interna­ an account of papers presented at SCCAC meetings and tional Conference of ISKO counts already as the 1st one additional contributions of members. There are also the of the German Chapter. abstracts from the "Leningrad MT Project". Under "Fu­ ture Activities" the programs of the 10th and IIth Inter­ First German ISKO Wo rking Group Established national SCCAC meetings are given, both held in Ger­ Upon invitation by Dr.Peter JAENECKE the WG on many (Siegen, June 5, 1990 in conjunction with the 17th 'Cognition and Kn owledge Organization ' was established ALLC and 10th ICCH Conference; and in Darmsiadt, on May 17, 1990 in Bonn. The members present elected Aug.14, 1990, in conjunction with the 1st International Mr.Jaenecke as its chairman. Its scope has been outlined ISKO Conference). Furthermore the Newsletter lists in ISKO News I already. A firstget-together is planned some upcoming conferences in 1990 and relevant publica­ to take place in Darmstadt, during the days oftheinterna­ tions, systematically arranged. It concludes with a list of tional conference to discuss this scope description as well SCCAC project reports (p.S4-60) and the membership as further work and meetings. list (p.61-78).

Int Classif. l7 (1990) No.2 -ISKO News J 101 logy and Knowledge Engineering Applications" in Trier, October 2-4, 1990; (4) The FID 45th Conference and Congress, Havana, FIDleR News 27 Cuba,September 17-22, 1990. (5) The ASIS Annual Conference, Toronto, Canada, NovA-8, 1990. ASIS has a variety of program meet­ Greetings after a rather long delay. My apologies. Un­ ings which focus on classification, thesauri, expert fo rtunately, I was unable to meet the deadline for the last systems and artificial intelligence, as well as a Special issue of In ternational Classification. However, I hope the Interest Group on Classification Research (SIGICR). "News" is now back on track. International Symposium on "Terminology in Advanced Task Force on UDC System Development Microcomputer Applications, Vienna, Nov. 22-23, 1989. The Task Force completed its work at a final meeting This International Symposium was organized by the held in Amsterdam in early February. The report was International Network for Terminology (TermNet) presented to the UDC Management Board in March, under the sponsorship ofInfoterm, and supported by the and was to go forward to the FID Council in April. InternationalInstitute for Terminology Research (IITF) Dr.I.McILWAINE, Chairperson of the Task Force will and Centre national de la recherche scientifiquejCentre be preparing a report which, hopefully, will be published de terminologie et neologie (CNRS/CTN). The Sympo­ in FIDICR News 28. Meanwhile, FIDICR members can sium was attended by more than 70 participants from 20 be assured that everything possible is being done to en­ countries. The aim ofthe Symposium was to present par­ sure the future ofUDC. ticipants with an overview of some ofthe microcomputer software available for the management of tenninology 5th International Study Conference on Classification Re­ as well as of their application in infOlIDation manage­ search ment. The opening address was given by Mr. Pierre LERAT, Vice President of TermNet and Director of the Plans are now well under way for the holding of this French Centre for Terminology and Neology. Mr. conference in Toronto, Canada, in June 1991. Tentative LERAT identified six difficulties facing terminologists, dates for the conference are June 23-28. Dates have been which he believed must be overcome in tenninology­ picked to permit those who wish to attend the American microcomputer applications. These were: The simulta­ Library Association Annual Conference in Atlanta, neous use of different alphabets, which is now possible; USA, June 29-July 4. The theme ofthe conference will be the semantic structure of concepts; inevitable differences "Classification Research in Knowledge Representation in concept systems, on national and international levels; and Organization". The Faculty ofLibrary and Informa­ the flexibility of computer processing systems; the port­ tion Science, University of Toronto will sponsor the con­ ability of programs and data and user-friendliness. Key­ ference and cosponsors are being sought. The conference note addresses were given by Professor Alan MELBY, will focus on classification, its theory, principles and ap­ Brigham Young University, Salt Lake City, United plications in the broadest sense, and will encompass re­ States, on "Terminology - an Indispensible Tool for All search into both universal classification and special sys­ Aspects of Information Management", and by Mr.Mi­ tems, as well as thesauri, knowledge representation and chael SCHAAR, InTra Fachiibersetzgenossenschaft "intelligent" systems. A formal, detailed, call for papers (Federation of Technical Translators), Stuttgart, will be sent out about June I, 1990. The deadline for the FRG.Programme presentations focused on a compari­ submission of abstracts will be October 1, 1990, notices son of traditional databases and dedicated termbanks; of acceptance of papers for presentation will be mailed the complexity of terminographic data exchange; do­ not later than January 2, 1991 and the deadline for com­ cumentation in computerized terminography; and the in­ pleted papers will be April I, 1991. A paper on "Classifi­ tegration of systems. At the conclusion of the Sympo­ cation Research since Augsburg" which will lead up to sium, the participants passed a recommendation that the conference is currently in preparation. since: "InfOlIDation management and knowledge process­ 1990 Conferences of Interest to FIDICR Members ing cannot do without tenninology and documentation, Highly recommended for FIDICR members and as well as without cooperation concerning exchange of those interested in classification research are the follow­ terminological data and experience, therefore: ing conferences: I. decision makers in industry, administrations and or­ (I) The ISKO Conference on "Tools for Knowledge Or­ ganizations are called upon to support and imple­ ganization and the Human Interface", Darmstadt, ment terminological approaches to information sys­ West Germany, August 14-17; tems in order to guarantee the accessibility of knowl­ (2) The 1FLA Annual Conference in Stockholm, edge by the different user groups in the appropriate Sweden, August 18-25, 1990, especially the program form at the time when needed; meeting of the Standing Committee on Classification 2. TennNetis requested to develop a common termino­ and Indexing; logical exchange format and make it available to the (3) The TKE'90 Conference of the Association for Ter­ general public. minology and Knowledge Transfer on "Terrmno- Nancy Williamson, FIDICR Chair

102 Int. Classif. 17 (1990) No.2- FIDleR News 27 U.K.: eRG 275 Barbara SNYDER from the Getty Conservation In­ The 275th meeting of the British Classification Re­ stitute (GCI) will give a talk titled "Conservation Infor­ search Group was held on Feb.15, 1990 at University Col­ mation Network (CIN) - Joint Development of the Ver­ lege London with 8 members present. The minutes of the non Toolkit, an ARev Application". In collaboration previous meeting were enlarged upon concerning the ela­ with Vernon Associates of Aukland, New Zealand, GCI boration of the preliminary classes AK and AY/AZ of has developed a general purpose data entry system for the Bliss Classification, 2nd ed., (regarding Systemo­ members of CIN which includes a sophisticated the­ logy, Management and Information Processing). The saurus construction module. The Vernon Toolkit is built main topic of this meeting were the problems of the BC on the Advanced Revelation (ARev) dbms and includes Classificationof Law. It had been agreed that the general the capability to create a hierarchically structured the­ arrangement of the subject would be: Jurisdiction - saurus which can be integrally linked to the data entry sys­ Legal Subject - Administration of Justice - Juris­ tem. prudence with the coverage of the theory of the subject SIG/CR program planners are pleased to have been and including comparative law. The schedule order will able to assemble some ofthe latest examples ofmicrocom­ reflect the filing order and will start out with Juris­ puter-based thesaurus construction systems. Since al­ prudence. The specific subclasses were also presented by most everybody today has access to an MS DOS or Ma­ Mr. Jacques MILLS and commented upon by the pres­ cintosh microcomputer, I often receive questions about ent members of the Group. The problem ofwhere should what packages are available for thesaurus construction special kinds oflaw be placed were among those to be dis­ on microcomputers. The best place to start is in the 2nd cussed, viz., ecclesiastical law, tribal 1aw and the law of edition of Thesaurus Construction by Aitchison and Gil­ specific jurisdiction, military law. Also of concern was christ, Section M "Packages and Service Bureaus". I the problem of the arrangement of national law. The have seen micro versionsof products from the following potential value offollowing Moy's arrangement underju­ vendors listed - ADLIB (Databasix, UK), Micro Stride risdiction was noted. - The next meeting was scheduled (BNF Metals, UK), and TINTERM (IME, UK). Micro for April 26, 1990 on the topic: "Problems in subject de­ versions of BASIS (Battelle), INQUIRE (Infodata), and scription in large online bibliographic data bases". MINISIS (IDRC) are rumored to exist but I have never seen them. And new software specifically developed for thesaural applications on microcomputers - Toolkit USA: SIG/CR Program for ASIS Mid-Year (Vernon) and TCS (Liu-Palmer). Joseph A.Busch (From ASIS SIG/CR News March 1990) The 19th ASIS Mid-Year Meeting will be held May 13-17 at the Bahia Mar Hotel in Fr.Lauderdale, Florida. The topic for the conference is "Microcomputing in the ASIS '90 SIGICR Program Preview 1990's: Unlocking the Power." The SIG/CR proposal for (From ASIS SIG/CR News March 1990) a session on thesaurus construction titled "Micro-The­ The ASIS'90 Annual Meeting will be held November sauri" has been accepted by the conference committee. 5-8 (1990) at the Sheraton Centre in Toronto, Canada. The "Micro-Thesauri" session will have presentations The SIG/CR program promises to be interesting and in­ about microcomputer-based thesaurus development formative, gathering speakers with a wealth of experi­ tools, as well as problems in the development and applica­ ence and knowledge and involved in projects at the fore­ tion of hierarchically structured controlled vocabularies. front of classificationresarch in information science. The following presentations are planned: A special part of our program is the CR Workshop to Ronald L.BUCHAN from NASA STI/RMS Associ­ be held Sunday, November 4. The Call for Papers ... (has ates will present a paper titled "Retrospective Indexing been published in Int.Classif.90-1, p.34 - Ed.). -What's new under the Sun", Mr.Buchan's paper is The following are proposals for SIG/CR Sessions de­ about computer-assisted techniques which facilitate veloped so far and presenters we have lined up. re-indexing following, for example, development of a new controlled vocabulary. Although the NASA prob­ Th esaurus Soft ware Sp ecifications lem is operating on larger computers, the machine-as­ SIG Sponsors: SIGs/Classification Research (CR) and sisted approach to the problem of re-indexing is com­ Personal Computers (PC). puter-independent. Kathleen MCDONNELL and Barbara SNYDER Description: Panel discussion of requirements, specifi­ from the Getty Conservation Institute (GCI) will give a cation, fe asibility, ease of use, effectiveness, efficiency, talk titled "Conservation Infonnation Network (CIN) and evaluation ofsoftware meant for thesaurus construc­ Structured Tools for Building and Accessing a Coopera­ tion. All participants have experience in both designing tively Developed Database". This presentation will dis­ and using such software. each will give a short presenta­ cuss various Gel Projects including development of a tion, with time for interaction and comment. controlled vocabulary for art and archaeology conserva­ Moderator: Raya FIDEL, University of Washington. tion, application of the controlled vocabulary retrospec­ Panelists: Dagobert SOERGEL, University of Mary­ tively to CIN databases, and development ofamicrocom­ land; Jessica L.MILSTEAD, Jelem Company; James puter-based system for standardizing CIN data collec­ D.ANDERSON, Rutgers University; 4th panelist to be tion in the future. announced.

Tnt. Classif. 17 (1990) No.2 -FIDleR News 27 103 Applications of Thesauri as Knowledge Bases fo r info r­ mat, identifiedby participants as follows: printing ofclassifi­ mation Retrieval cation schedules, browsing, access to infonnationinc1assifi­ SIG Sponsors: SIGs/Classification Research (CR) and caiton records, authority control, machine-assisted classifi­ Human-Computer Interaction (HCI). cation, expert systems, conversion from one classification Description: While the traditional application of the­ system to another, and a consolidated index to theclassifica­ sauri is for vocabulary control for indexers and sear­ tion (e.g., index to all LCC schedules). chers, a thesaurus can also be viewed as a knowledge Treatment ofthe classificationnumber and its caption base, that is, a compilation ofconcepts connected by rela­ hierarchy was discussed. In the current version of the Pro­ tions. As such, thesauri can be used effectively in a num­ posed Classification Format, the complete caption hier­ ber of knowledge-based applications. Important applica­ archy accompanies the classification number in the 153 tions to be discussed in this session are intelligent inter­ field(classification number field).Inclusion ofthis infor­ faces for retrieval and indexing and networking of dis­ mation in the same field as the number remains an unre­ parate information systems. solved issue, as does the more general problem of how to Moderator: Barbary H.KW ASNIK, Syracuse Univer­ link numbers where the notation is not expressive of the sity. Presenters: Alina VICKERY, Tome Associates; hierarchy. Deirdre C.sTAM, Syracuse University; Jennifer A separate document, Discussion Paper 33: Ref­ A.YOUNGER, University of Wisconsin and Prudence erences in the Classification Format, was discussed in W.DALRYMPLE, University of Illinois at Urbana; conjunction with treatment of 'see' and 'see also' ref­ Trish YANCEY, Information Handling Services. erences in the format. Gary STRAWN of Northwestern Applications of Classifications University studied earlier versions of the format and pro­ SIG Sponsor: SIG/ClassificationResearch (CR). posed that references be treated in a way analogous to Description: Topics include extending and improving treatment of references in the authority format in 4xx use of existing classifications, methods and problems in and 5xx fields. MARBI agreed to pursue the idea, but developing classification schemes· for document collec­ notes that there are many special circumstances (es­ tions, and proposals for use of classification in auto­ pecially in the case of some Dewey notes) which may re­ mated text analysis. quire double coding. Moderator: John S.MITCHELL, Carnegie-Mellon Participants agreed that the handling ofinternal tables University. Presenters: Nancy J.WILLIAMSON, was problematic. A suggestion was put forth to separate University of Toronto; Clare BEGHTOL, University of 7xx fields into two blocks: 74x (Access fields) and 76x Toronto; Mary DYKSTRA, Dalhousie University; Mar­ (Number building fields). In the latter, a subfieldcontain­ lene A.PALMER, Farm Credit Administration. ing the appropriate field number from the format might Barbara H.Kwasnik be added to handle the problem of ordering elements (e.g., notes) found within the internal table. The Proposed Classification Format generated much Classification News discussion, but no formal action was taken by the com­ (From ASIS SIG/CR News, March 1990) mittee. It will be revised again this Spring and discussed Proposed Classification Fo rmat at the MARBI meetings during the ALA Annual Con­ Proposal 89-1, Proposed MARC Classification For­ fe rence in Chicago at the end of June. mat, was reviewed as a discussion paper at meetings of Subject Code the American Library Association Machine Readable The Cataloging and Classification Section of the ALA Bibliographic Information (MARBI) Connnittee during Association for Library Collections and Technical Ser­ the ALA Midwinter Meeting in Chicago in early vices will sponsor a program, The Subject Code, at the January. Topics discussed included uses of the format; ALA Annual Conference, Chicago, June 24, 1990, treatment of numbers, captions, and hierarchies; hand­ 2:00-5:30 M. Speakers and topics include: ling of 'see' and 'see also' references; extensive suhfield­ Wh at is a Subject Code?" Mary DYKSTRA, Direc­ ing for data identification;and accommodation oftables tor, School of Library and Information Studies, Dal­ and index terms. house University. Sally MCCALLUM, Library of Congress Network Do We Have One? Do We Need One? William Development/MARC Standards Office, introduced the E.STUDWELL, Principal Cataloger, Northern Illinois format by reminding MARBI and observers that the University. prime purpose of any MARC format is communica­ Models fo r Subject Code Development and Govem­ tions. In the case of the Proposed ClassificationFormat, ance: Dewey Decimal Classification. Liz BISHOFF, secondary goals include providing enough content for Manager, Cataloging and Database Services, OCLC. browsing, filemaintenance, and data for automated ap­ Models fo r Subject Code Development and Govem­ plications of classification. ance: AACR2R. Michael GORMAN, Director, Univer­ Rebecca GUENTHER, also of Library of Congress sity Libraries, California State University, Fresno. Network Development/MARC Standards Office, intro­ Communications regarding this report should go to duced the format and asked the group to focus on prin­ Joan S.Mitchell, Director, Academic Computing & In­ ciples for the different blocks of fields, i.e. fixed fields, structional Technology, Carnegie-Mellon University, classification number, references and tracings, note 5000 Forbes Ave., Pittsburgh,PA 15213-3890, USA. fields, etc. Discussion began with potential uses ofthe for- Joan S.Mitchell

104 Int. Classif. 17 (1990) No.2- FIDleR News 27 fo rmation problems (verificative, topical muddled, con­ scious with label, and conscious with no label), and then proposing some design solutions. Book Reviews Brigitte ENORES-NIGGEMEYER uses the Kintschf van Oijk model of text comprehension as a vehicle for bet­ KOSKIALA, Sinikka; LAUNO, Ritva (Eds.): Informa­ ter content analysis. Because she believes that "our tion - Knowledge - Evolution. Proceedings of the 44th knowledge about abstract writing fits into a general writ­ FID Congress, Helsinki, Finland, Aug.28 - 1 September, ing model", she uses this model after elaborating the task 1988.Amsterdam, NL: North-Holland 1989. 466 p. environment. The result looks very much like the "seman­ ISBN 0-444-87326-0; FID-Pub1.675 tic differential" ofgeneral semantics theory in the 1950's. TheInternational Federation for Information and Do­ Examples oftext reduction, condensation, and reorganiz­ cumentation (FlO) againhas spanned the specialized dis­ ation (following macrarules and substracting rules) are ciplines of librarianship, documentation and informa­ interesting, but as she concluded, and we concur, "it is a tion science by holding a conference where topics com­ long way to go from the model of content analysis mon to all three - classification, data processing, com­ sketched above to a fully-fledged model that defines the munications, and technology transfer - could be dis­ intellectual process of content analysis with the desirable cussed and debated. The handomly bound and printed precision and reliability". volume contains papers presented at eight sessions, rang­ "Retrieval differen'ces between term and citation in­ ing from "Man as an Information Processor" to "Elec­ dexing", is a straightforward progress report of a small tronic Knowledge" and "National and International In­ scale analysis by Miranda Lee PAO. The paper fram formation Policies", This review will concentrate on the Lesotho by M.M.MOSHOESHOW, "Content analysis ten papers presented during Session 3 (pages 77-181): and description of documents" also falls into this ca­ Content Analysis and Description of Documents, and tegory of " progress report", but this time using the case two other papers which would have considerable interest study method observing four major information services for the readers of this journal. since the 1970's. It provides a rare glimpse into such work Amy J.WARNER's paper, "Linguistic theories for in­ in a developing country. formation retrieval", lives up to the prediction in the Papers fr om West Germany, Cuba, and Japan high­ opening address of Michael W.Hill, President of FlO: light efforts to handle free text databases (G.RUGE and " ... by devating and discussing opinions ... eliminate the S.SCHWARZ, "Natural language access to free-text da­ bad ones, improve the soundly based ones and shape the tabases", "Automatic indexing in Spanish language of good ones into hypotheses". She revisits the terrain writ­ Russian scientificand technical texts", by Carmen I.CA­ ten about since the mid-60s, then the theme of several ZARES, and "A dynamic thesaurus for intelligent access papers by Christine MONTGOMERY (not cited), to research databases", by Y.FUJIWARA, et a\.). Karen SPARCK JONES, and J.C.GARDIN (cited), The tenth paper to be reviewed reports on the reorgan­ alerting a new audience to the need for more interdiscipli­ ization of the process to handle UOC revision. E.SCI­ nary work to reorient some of the infOlmation retrieval BOR and I.SHCHERBlNA-SAMOJLOVA describe field'sassumptions ofprocedures concerning indexing lan­ the old process first and then the new one which encom­ guages and the structure of texts. Her plea is grounded in passes several facets and aspects which Alan GIL­ the issues being faced by research and application in the CHRIST and Partners recommended in their 1984 man­ field of natural language processing, namely: the cognitive agement study. versus engineering approaches; the units and components Two other papers, not in Session 3, deserve some no­ of data analyzed; and the breadth of subject domain. This tice. "Wholly new forms of encyclopedias: Electronic is a very provocative and timely paper. knowledge in the form ofhypertext" by Linda C.SMlTH Irmeli HOV!'s paper, "The cognitive structure ofclasi­ (pages 245-250) reviews the need for redefining the fea­ fication work", is provocative in another way. By stu­ tures of the encyclopedia to avoid disorientation, to pro­ dying the UOC and LO (Luokitus opas -Finnish adapta­ duce maps or graphical browsers -an area where classifi­ tion of Dewey) classification work of26 librarians and 9 cationists may be of some help. She concludes by saying, students of library science by the thinking aloud method, "Sequence need not be limited to a single alphabetical ar­ several conclusions could be drawn about the expression rangement, for there can be mUltiple paths defined... The of concepts in classification systems and the influence cross reference structure can be much richer, and more this has on the analysis of the classifiers handling books easily traversed ... Given the many infonnation organiza­ in the social sciences, and the inconsistencies in their tion and access questions to be addressed, librarians and classification codes. Not a definitive piece of research, infonnationscientists have much to contribute to this en­ but very interesting as classification systems go online as deavor". Having seen the CD-ROM version of The indexing and retrieval tools. WOrldBook Encyclopedia, with a split screen showing the "Modern indexing and retrieval techniques matching article outline and the text, with indexed words high­ different types of information needs", by Peter ING­ lighted, this reviewer knows a new argument can be made WERSEN and Irene WORMELL provides a useful re­ for more indexers and lexicographers working closely view of retrieval techniques (exact match, partial match, with encyclopedia editors and graphic artists. probabilistic models and clustering techniques, and "A blueprint of an intermediary system for numeric Boolean logic), juxtaposing these techniques against in- source databases", by Kalervo JARVELIN (pages

Int. Classif. 17 (1990) No.2 - Book Reviews 105 311-320) will be interesting reading for anyone interested The book divides into 13 chapters. Following two short in multiple bibliographic database vocabulary switching introductory chapters, Buchanan explains the basic dif­ because it "analyses the functional requirements of deal­ ferences between precombined and facet classificationsys­ ing with NDBs in distributed environments, where the tems. He then leaves precombined systems at what they are data models used to organize the databases vary, where and concentrates on facet classification.Yet in spite of the the data representations are not directly compatible, and book's title, Buchanan's approach is not at all theoretic. where the naming of data item types is not standardized He rather tends to the pragmaticside and documents, very resulting in difficulties in relating data from different often in minute detail, the design of an exemplary facet sources". As these are common problems in bibliograhic classification.The chosen example is zoology. Which may databases as well, this analysis can offer some assistance not be everybody's cup of tea. However, zoology lends it­ in that area too. self to such a demonstrational purpose, being unbiased by The editors performed a remarkable feat in getting this any ideological slant and multilingual, as it were: for we all volume published within the year after the meeting, mak­ know what a butterfly is, don't we? That explanation takes ing the reading of the papers more lively and current. Al­ about half of the book, with due attention being paid to though the cost would appear to be exorbitant, even for matters such as tenninology and itsstructuring, notation, the most affluent library, it still needs to be made avail­ or alphabetic index. able whereever there are students and researchers in the In order to popularize facet classification in German field oflibrary and infonnationstudies. speaking countries, though, the exclusive realm of zoology Pauline A.Cochrane should have been extended to include more examples, not­ ably from disciplines concentrating on conceptual systems more abstract than zoology. What about matters such as BUCHANAN, Brian: Bibliothekarische Klassifikation­ philosophy, history, politics, economy? To show that facet stheorie. Miinchen: K.G.Saur Ver1. l989. 151p. classification is actually a feasible approach with disci­ ISBN 3-598-10788-9 plines involving ideas rather than things would have been tremendously helpful. Ten years after the publication of Buchanan's Theory of The translation preserves what may be regarded the Library Classification(London: Bingley 1979) it is good to most important fe ature of the original, its clear and un­ have this book translated. At last, one may add. For the pretentious style. The steps to be taken in designing a facet slim introduction into the intricate business of classifica­ classificationare laid out in great detail, making the whole tion has made its way to the textbook shelves for students process transparent. Occasional references to otherclassifi­ of librarianship in English speaking countries. And al­ cation approaches (notably critical remarks on the ODC) though similar progress is desired for the translation, there notwithstanding, the translation should succeed in hold­ are some reservationss. ingitsreaders to its course very much the same way the orig­ For two reasons, Buchanan's book should meet a heavy inal version does. And that course is nothing more (and in­ demand in this country. First, surprisingly enough, there is deed nothing less) than a plain description of designing a hardly any introduction or textbook on classificationthat facet classification. As for matters of classification tenni­ would serve the German speaking and classifying market; nology itself, the translation provides helpful references to the last attempt was, according to the majority of reviews, Buchanan's original text as well as to Gennan DIN stand­ nothing more than precisely that (I). Second, and what is ards. What remains to be criticized, if rather fonnally more, Buchanan's book paves the way to an appreciation though, is the quality of the book itself: the word-process­ of facet classification, an approach which has alinost al­ ing layout, the fatiguing print-area, the offset-printing as ways been considered somewhat exotic in traditional Ger­ well as the soft binding do not exactly look like demanding man classification theory and practice. The only book on DM 48. Moreover, the translation renders exactly the orig­ facet classification for Gennan speaking readers before inal version published in 1979. Ten years later, references the present translation of Buchanan's book was published should have been updated, and students of librarianship twenty years ago (2). should no longer be referred to the 18th edition of ODC. Yet for all book's virtues, it should be stated that some As for the title of the book, we cannot blame the publisher virtues are only pretended. A major reservation is the title that the original version was preserved; something like a of the book. It simply promises too much. Buchanan does "demonstration of facet classification" would, after all, not provide an introduction, let alone an outline of library hardly meet anything more than a minority's interest in classification. The book is rather a straightforeward, if ex­ this country. Heiner Schnelling clusive, introduction into a pragmatic approach to facet classification. Even with respect to English speaking coun­ (1) Weisshaupt, K.: SacherschlieBung in Bibliotheken und Bib­ tries and their tradition of facet classification, any survey liographicn I: Klassifikatorische SacherschlieBung: Frank­ of library classificationwould have to be considered incom­ furt: Klostermann 1985. (Among the numerous reVIews, sec W.Godert in Int.Classif. 13(1986)p.39AO, G.Heinrich in plete that dealt only with facet classification and left only Bibliothek. Forsch. u.Praxis lO(l986)p. 136-139) some introductory remarks to precombined ones, notably (2) Vickery, B.C.: FaceHenklassifikation. Pullach b.Miinchen: DOC. It is not surprising, therefore, that Buchanan's book VerLDokumentation 1969. (3) Foskett, A.C.: The subject approach to infonnation. 4th ed. is marked "elementary" in Foskett's seminal textbook (3). London: Bingley 1982. p.ll That verdict may also draw on the misleading title, for the book is hardly theoretic. On the contrary, it is downright Dr.H.Schnelling, UniversiHitsbibliothek Giesscn, Otto�Be� pragmatic. haghel-Str.8, D�6300 Giessen

106 Tnt. Classir. 17 (1990) No.2 ·· Book Reviews COMAROMI John P. (Ed.), Beall, J., Matthews, Leader". In 1986, Lois Mai CHAN took over as chairper­ W.E.,Jr., New, G.R. (Assist.Eds.): Dewey Decimal son of the DCEPC from Margaret E.COCKSHUTT of Classification and Relative Index. Devised by Melvi! Toronto whose tenure had been from 1980 to 1985. Dewey. Ed.20. 4 vols. Albany, N.Y. Forest Press. OCLC 1989. LXi,514 + XX,944 + VII,864 + VII,96Ip. As another far reaching change, for the firsttime in li­ ISBN 0-910608-37-7 brary history, a library classificationhas been edited and A new edition of a classification adopted world wide produced on a micro computer system called - in this by 200,000 cities in 135 countries is always a celebrated case - Editorial Support System (ESS) installed in 1986; and exciting event. A new edition ofthe DOC has always and a DOC database has been created. It is the firstedi­ created a bit of a stir in the library profession and the li­ tion generated from an online database. Programming brary press. It is a result of intensive and wide deliber­ was done by John J.FINNI and Cora M.ARSENAULT ations and work of ten years by classificationspeciali sts, from Inforonics, Inc. (USA), commissioned by Forest teachers, library users and expert committees fr om USA, Press in 1984. ESS has come as a great help in editorial re­ Canada, and the UK. visions. Revisions, additions, deletions, relocations of numbers and headings can now be done instantly. The DDC-20 released in January 1989 comprises 3378 ability to construct, edit, modify, and develop classifica­ pages in four volumes. The schedules earlier in va1.2 have tion systems through interaction with a computer data­ now been put into two vols. (2-3) for the convenience of base has many advantages. It makes possible the rapid physical handling and ease of referring to - otherwise finding of all uses of any DOC number or its verbal de­ there is nothing of unusual increase on the part of the scription from the schedules or any of the seven tables. A schedules. However, DDC-20 brings in many more major advantage is the ease and rapidity of printing draft changes compared with the last two editions (18 and 19) schedules for review by the editorial staff and for published in 1971 and 1979 respectively. Outstanding DCEPC, and for finalprinting. It will haveenonnous im­ changes with many implications and potentialities for plications on publications, namely frequency, cost and trend setting have occured in its administration. Since fo rmat. From the publisher'S point of view the classifica­ August 1988, its publishers, the Forest Press at Albany, a tionis now always in standing type readily amenable to al­ subsidiary of the Lake Placid Club Educational Founda­ ternations of any magnitude here and there: always min­ tion established by Melvil Dewey in 1896, has been utely up-to-date and always ready for timely publica­ owned by and incorporated into the famous OCLC On­ tion. By this way the time for its physical production has line Computer Library Centre at Dublin, Ohio. It marks been reduced to one-fourth only. Soon the system will be a new era for its testing and use in online retrieval systems made available online and in CD-ROM versions. A begun by the research work of Dr.Karen Markey MARC authority format is being actively developed at DRABENSTOTT. As another change, it is the firstedi­ the Library of Congress. tion produced by a new editor, DrJohnP.COMAROMI who took over in 1980 from Benjamin A.CUSTER who In the new edition, as usual, the changes in contents are retired after editing four (16-19) highly successful edi­ concerning expansions, deletions, and relocations of old tions fr om 1956 to 1979. J.P.Comaromi has been associ­ classes, and additions ofafewnew ones. As a majordevel­ ated with the DDe since long in various capacities: his opment a completely new schedule (DDC-20 avoids the doctoral research on the history of the DOC completed term phoenix schedule) for 780 Music as been incorpor­ in 1969 at Michigan University and his serving as the ated. This schedule had been prepared outside the Deci­ chairperson of the Decimal Classification Editorial Pol­ mal Classification Division of the LC by British experts icy Committee (DCEPC) from 1976 to 1979. By inclina­ in 1980 and had been widely circulated for opinions of tion a philosopher and a scholar, Comaromi believes in users and teachers before its finalinclusion. The process constant and drastic revision of the system to keep equal took ten years. It portends a welcome trend in broad­ pace with growth and changing structure of knowledge. based revisions - a laudable policy which seems to have However, the new edition toeing the line of the preceding been borrowed from the UDC. Other major revisions of edition maintains, as if diplomatically, a delicate balance 570 Life Sciences and 370 Education will now probably between stability and need to change. be included in DDC-21. The 20th edition also incorpor­ As another administrative change, the Editor of the ates schedules for 004-006 Data Processing and Com­ DOC, until December 1986 also headed the Decimal puter Science, and 301-307 Sociology, both schedules al­ Classification Division (DCD) of the Library of Con­ ready issued as separates since the publication ofDDC- gress. It is at this office where the DOC is constantly ap­ 19. Minor changes have taken place for class numbers of plied to ahnost 100,000 books annually, and changes are Christian Religion, Television, Adult Education, Elec­ formulated and tested by the DCD, the laboratory of the tronics, Civil Rights, Gymnastics, and Central American DOC being at the same time the single largest user of this History. However, no special efforts have been made to system. Since January 1987, headship of the Division reduce the American and Christian bias in the system. and editorship of the DOC have been separated. David Geographic area numbers have been modernised corre­ A. SMITH is now the head of the Division. In June 1985, sponding to the political and administrative changes Peter J.PAULSON succeeded John A.HUMPHRY as throughout the world since the last decade. Accordingly Executive Director of Forest Press. As a great tribute to changes have been made in the area numbers for British his work, this 20th edition has been dedicated to John Columbia, Japan, Nigeria, South Africa, Melanesia, Ames Humphry with the citation "Gentleman and Papua New Guinea, Switzerland, Germany and Belgium.

Int. Classif. 17 (1990) No.2- Book Reviews 107 Some of the important changes have occurred in the ss, and area in the index of DDC-19 are now replaced by other tables too, especially in Table 3, Subdivisions for In­ T1 and T2 respectively. The advantage is obvious. dividual Literatures ... (p.389-414, Vol.l). Now, instead The new edition on permanent paper and in maroon of two, the DDC has three subtables: 3-A, 3-B, and 3-C, buckam binding is attractive to look at and compara­ the latter being in fact a subsubtable as it is mostly to be tively easy to use. The size of the pages has been a bit re­ used when instructed in Table 3-B and to some extent duced by lessening margins on all sides thus leaving little also in 808-809. This further fu rcation makes its applica­ space for personal notes on the margin. Many librarians tions direct and simplified. will regret this. I.Comaromi has furthered the tradition Format and presentation of the text has been im­ of innovation and user's convenience set by the illust­ proved in elegance and operation. The three major sum­ rious predecessor B.A. Custer, now Editor Emeritus. maries have been relocated to the second volume. More­ The DDC will play its role also in the information so­ over, throughout the text there are more multilevel sum­ ciety of the 21st century with envisaged less paper li­ maries especially in dense schedules such as 620 Engineer­ braries. Maybe, DDC-21 is amongst the firstfew import­ ing' 630 Agriculture, 370 Education. Multilevel sum­ ant events of the library and information world at the be­ maries are provided for eight major divisions and for the ginning of the 21st century. In that case the symbolism Area Table for Europe and North America. With such may not merely be a coincidence. It will reflect its mani­ multilevel summaries, the structure of the entire class can fested urge to live through adaptability, and to serve the be seen at one glance. Inlocatingnumbers, this is time sav­ users by embracing needed modernization. ing, and subjects the schedules to a less flipping of pages M.P.Satija and thus to less Wear and tear. Under each entry copious definitional, scope and in­ Mohinder Partap Satija, Guru Nanak Dev University, Depart­ ment of Library and Information Science, Amritsar-143 005, structional notes have been appended. Centered head­ India ings are now flagged typographically by the symbol in the number column. Optional numbers are given in parentheses, e.g., Indian English literature (828.9935).

The Manual on the Use of the DDC (Forest Press, SHARMA, Pandey S.K.: Universe of Knowledge and Re­ 1982) has been incorporated into the DDC (Vol.4, search Methodology. Delhi: Ken Publication 1990. p. 731-968) and is referred to at the appropriate headings. XII,165p. Themanual is the blue book ofthe policies andinterpreta­ tions of the DDC at the Decimal Classification Division. Knowledge is librarian's stock-in-trade. The study of the With its many maps and flow charts it is a unique and nature of knowledge is as important to an information laudable feature of the DDC. Also, it comes as a handy scientist as the study of anatomy to a surgeon. Its implica­ guide to the classifiers. Since in every classification sys­ tions in information handling are too numerous. tem, class numbers are liable to varying interpretations, S.R.Ranganathan is a pioneer in the studies on modes of such an official manual will promote uniformity in the knowledge growth. In the year 1948 Ranganathan de­ use of the system in different libraries throughout the veloped and got introduced a course entitled "Develop­ world. ments and Structure of the Universe of Subjects" in the In order to meet the long standing demand of the users master degree curriculum of the University of De1hi. He especially outside the English speaking world, the Edi­ had an abiding interest in this field, and always delved tor's introduction has notably been made simple and deep into it obtaining fr esh results. His announced book brief. "This Introduction is written primarily for the no­ on the subject, however, was never published. The work vice or beginning classifiers, although the experienced has been continued by his schoolmen at DRTC, Banga­ classifier may benefit from reviewing its contents" (Intro­ lore and elsewhere. The late Professor Jesse H.Shera duction, 1.1). The entire Editor's Introduction (I, (1903-1982) lauds this as Ranganathan's everlasting "in­ p.XXV-l) has been divided into small sections with fea­ tellectual contribution to the underlying philosophy ofli­ ture headings and section numbers. It has added to the brarianship"'. Ranganathan still dominates the Indian li­ convenience of referring to the instructions. brary school curricula both in contents and influence.Al­ Its index, an integral part of the system, and acknowl­ most all Indian library schools have introduced the paper edged as Dewey's other, innovating contribution to li­ "Universe of Knowledge" clubbed with "Research Meth­ brary science, has now been substantially trimmed fr om odology" to make out of this a full length course for the 1216 pages in DDC-19 to only 730 in DDC-20. This has master degree curriculum. Over the years the spirit and been accomplished without imparing its efficiency. It is a objectives of the paper have got clouded and many col­ result of intensive research that had been undertaken by leagues now disparagedly differ on the contents of this the editor and his team on the use and retrieval efficiency paper. Student do not know what exactly they are to of the relative index of the DDC. All "see" references learn; what sort of questions will be put to them; and, have been replaced by direct entries, e.g., by looking at what will subsequently help them in their information both "birds" and "aves" one reaches 598. Also, syn­ work. Consequently. four books 2-5 written on the sub­ thesised number abundantly present in the DDC-19 ject in India vary from history of scholarship to the his­ index, have been axed restricting it as an index to tennino­ tory and growth ofvarious disciplines, even to the anthro­ logy of the schedules. Abbreviations occurring in all of pological evolution of man and culture since prehistoric the seven tables have been replaced by their number, e.g., times. The situation has not become very clear despite the

108 Int. Classif. 17 (1990) No.2 - Book Reviews deliberations of a national seminar on the teaching of the man Jawaharlal Nehru (1889-1964). This chapter gives Universe of Knowledge held in 1986 at Hyderabad. Inter­ an overview ofthe major research institutions and the in­ pretations of the contents of this paper vary and some of fo rmation systems to feed them (p.133-136). Reviewing the contents are irrelevant for library trainees. the purpose of research in India, the author laudingly as­ cribes it to the pursuit of Truth, Welfare, and Beauty. Dr.SHARMA's textbook provides the students with Here, he ignores India mindlessly joining the rat race to some more relevant readings. It has been written strictly explore space and venturing into high-tech war weapons from the student viewpoint. As the title suggests, it is a bi­ while some hundred million of illiterate and hungry In­ focal book. The chapters AlE are on the Universe of dians look with a resigned fate towards GOD for their Knowledge, its definition, scope, nature, relations with daily subsistence. Nor talks he of the poor quality of In­ other subjects, and the modes of knowledge growth. An dian research despite the fact that India has the third lar­ important chapter fo cusses on number, order, and scope gest R&D manpower in the world. There are so many fac­ of main classes in UDe, DDC, and Colon Classification tors which militate against creating an embracing envi­ systems. A serious omission is the Bliss Classification ronment for research workers in India. It is time to pon­ and to some extent the Library of Congress Classifica­ der: Why so much of it and so diluted? tion. It provides more of a cut and dried curriculum read­ The text closes with two model research designs both ing than any scholarly and comparative study on the na­ on library problems. One is the author's own Ph.D. re­ ture of main classes and on the inherent relations be­ search design. It may be of practical help to the prospec­ tween the conceptual classification and the universe of tive doctoral students of Indian universitities. The book subjects. The author simply rehashes Ranganathan's is usefulfor Indian students and expecxtedly it will find fa­ views in a lucid and simple language. The book having vour with them. Mohinder Partap Satija arisen from a series of extension lectures to library science students, the author is successful now in com­ municating with the wider audience. Even examples are References homely. Diagrams and illustrations further help to com­ 1 Shera, 1.R.: Sociological foundations of librarianship. Bom­ prehend the text. bay: Asia 1962. p.106-107 2 Aggarwal, D.S.: Lectures on universe of knowledge. Delhi: The other part of the book (chapters G/H) is devoted Academic Publications 1985. 166p. to Research Methodology. It explains types of research 3 Khanna, 1.K., Vashisth, K.K.: Knowledge: Evolution, struc­ with brief explanations of each method. It concludes ture and research methodology. New Delhi: Ess Ess 1985. 519p. 4. Rajan Pillai, c.¥.: Universe of knowledge: nature and dimen­ with a valuable chapter on the organization of research sions. Trivandrum: Velicham Publications 1983. 64p. in India. About 95% of the R&D activities in India are 5 Shanna, 1.S.: Knowledge: its origin and growth from earliest government fu nded. As back as 1958, the Indian Parlia­ times to the present. New Delhi: Sterling 1978. 156 p. ment adopted the Science Policy Resolution on the initia­ M.P.Satija, Guru Nanak Dev University, Department of Li­ tive of India's first PrimeMinister and Visionary States- brary and Information Science, Amritsar-143 005, India

To wards the Millenium. 2nd Summer School for Information Science, Konstanz Library Catalogues for the Nineties The University of Konstanz Department for Informa­ The Cataloguing and Indexing Group of the (British) tion Science announced its "2.Sommerschule der Infor­ Library Association will hold its Annual General Meet­ mationswissenschaft", to be held in Konstanz from 3-7 ing combined with a Residential Weekend Seminar at Sept.J990. It is organized in collaboration with the Ge­ Pollock Halls, University of Edinburgh, July 13-16, sellschaft fUr angewandte Informationswissenschaft 1990. The keynote paper will be given by Peter e.V. Its professors are: Dr. Rainer KUHLEN, Dr.Her­ BROPHY, Lancashire Polytechnic on "General trends bert STOYAN, Dr.Ulrich REIMER, and Mr.Peter and prospects for the nineties". The six fu rther papers Dambon, Dipl.Inform. There are courses on Knowledge will be: (Speaker from UK Office for Library Network­ Representation; Introduction into logic programming ing): Network systems in the US. - Henry NEANEY, with PROLOG; Hypertext: Design principles and appli­ Glasgow University Libtrary: A UK national biblio­ cations; and Introduction into object-oriented pro­ graphic network. - Janet MITCHELL, OCLC Europe: gramming exemplified by SmallTalk. -For further infor­ Who will provide a national bibliographic network? - mation contact Mr .Beck, Sekretariat Infonnationswis­ Douglas ANDERSON: Standards for access in the senschaften, UniversitiitKonstanz, Postfach 5560, 7750 OPAC age. -Jack MEADOWS, Loughborough Univer­ Konstanz 1. sity: The Wall comes tumbling down - and the electronic barriers are raised. - (Speaker to be confirmed): Euro­ pean developments. - For further information turn to Ms.Mabeth CURRY, Database Management, Edin­ burgh City Libraries, Central Library, George IV Bridge, Edinburgh EHI lEG, U.K.

Int. Classif. 17 (1990) No.2 - Book Reviews 109 Classification Literature 1 '7(1 990)No.2 (2-001 - 2-356)

0 FORM DIVISIONS 5 ON SPECIAL OBJECTS CS (TAXONOMIES)'

01 Bibliographies 111 51 On Taxonomies In the Form & Strucl.Area 1 02 Literature Reviews 111 52 On Taxonomies In the Energy & Matter Area 2 03 Dlctfonarles. Terminologies 111 63 On Taxonomies In Cosmos & Earth Area 3 04 Classlf, Systems & Thesauri· 111 54 On Taxonomies In the Blo Area 4 05 Periodicals and Serials 112 55 On Taxonomies In the Human Area 5 06 Conference Reports, Proceedings 112 56 On Taxonomies In the SocIa Area 6 07 Text Book.s·· 113 57 On Taxonomies In the Prod.Area 7 08 Other Monographs·· 113 58 On Taxonomies In the Sd.& Info Area 8 09 Standards, Guidelines 113 59 On Taxonomies In the Culture Area 9

1 THEORETICAL FOUNDATIONS 6 ON SPECIAL SUBJECTS CS & T'

11 Order and Classification 61 On CS & T In the Form & Struct.Area 1 119 12 Conceptology In Classification 113 62 On CS & T In the Energy & Matter Area 2 119 13 Mathematics In Classification 114 63 On CS & T In the Cosmos & Earth Area 3 119 14 Systems Theory In Classification 64 On CS & T In the Blo Area 4 119 15 Psychology and Classification 114 65 On CS & T In the Human Area 5 119 16 Developm�nt of Science and Classll. 114 66 On CS & T In the SocloArea 6 119 17 Problems In Classification 114 67 On CS & T In the Productlon Area 7 119 18 Classification Research 68 On CS & T In theSci & Inlo Area 8 19 History of Classification 114 69 On CS & T In the Humanities Area 9 119

2 STRUCTURE & CONSTRUCTION OF CS & T'" 7 CLASSIFICATION AND LANGUAGE

21 General Questions of CS & T 114 71 General Problems of Natural Language 120 22 Elements of CS & T 114 72 Semantics 120 23 Construction of CS & T 115 73 Automatic Language Processing 120 24 Relationships 115 74 Grammar Problems 121 25 Numerical Taxonomy 115 75 Online Problems and Tedmologles 121 26 Notation, Codes 115 76 lexlcontDlctlonary Problems 123 27 Revision, Updating of CS & T 77 General Problems of Terminology 123 28 Compatibility & Concordance 115 78 SUbJect-Oriented Terminology Work· 123 29 Evaluation 01 CS & T 115 79 Problems of Multilingual Systems 124

3 CLASSING AND INDEXING 8 APPLIED CLASSING AND INDEXING (C&I)'"

31 Theory of Classing and Indexing 115 81 General Problems of Applied C&r 124 32 Subject Analysis 116 82 Data Classing and Indexing· 33 Classing and Indexing Techniques 116 83 Title Classing and Indexing 34 Automatic Classing and IndexIng 116 84 C&I of Primary Literature 124 35 Manual & Automatic Ordering Techn. 116 85 (Back of the) Book Indexing 124 36 Coding 66 Indexing of Secondary Literature 124 37 Reclassification 87 C&I of Non-Book Materials 124 36 Index Generation and Programmes 117 68 C&I In Subject Fields· 124 39 Evaluation 01 ClaSSing & Indexing 117 89 Free

4 ON UNIVERSAL SYSTEMS 9 CLASSIACATION ENVIRONMENT

41 On Universal Systems In General 91 Organization Problems In General 42 On the Universal Decimal Classlf. 118 92 Persons and Institutions 125 43 On the Dewey Decimal Classlf. 118 93 Organlz.of C&I on a Natl.& Int.Level 125 44 On the L1br.of Congress Classl!. 118 94 Free 45 On the Bliss Blbllogr.Classif. 95 Education and Training In C&I 125 46 On the Colon Classification 118 96 Legal Aspects of C&I 47 On the L1br. Blbllogr. Classl!. 118 97 Economic Aspects of C&I 48 On Other Universal Systems 119 98 User Studies 125 49 Free 99 Standardization in C&I 125

Personal Author Index 125

• Divisions of 04, 5, 6, 78, 82 and 88 according to the Information Coding Classification (ICC) desC(lbed and displayed In Int.Classlf.9(1982) No.2, p.87-93 and In the International ClassJllcatlon and Indexing Bibliography, We offer to send copies of the abstracts of the references cited to anybody Vol .1.(1982). Interested.

A further division of the above scheme (by one and sometimes two levels) was published In Int.Classlf.12(1 985) No.3, p.147-151. Language abbrevIations; From thIs more detailed scheme II will bevisible that subject fields may also Bulgarian FInnish pi Polish occur In many cases on the third and also forth level of divIsIon, Indicated by bg II fr French pt Portuguese the number 8. cs Czech da Danish hu HungarIan ro Romanian de German It Italian ru Russian .. Monographs on special topics at spadal subdivisions. ee Eston1an la Japanese sh Serbo-Croatian en English II LithuanIan sk Slovakian 'U CS & T stands for Classification Systems and Thesauri es Spanish nl Dutch uk Ukrainian C & I stands for Classing and Indexing

110 Int. Class if. 17(1990)No.2 - Classification Literature glk, 5.Aull. (Classification Scheme D: Eduqatlon, 5th ed,) (Orlg. de). Re­ o Form Division gensburg, DE: Unlv.8lbllothek 1987. 54p.

2-010 041-de-924 01 Bibliographies UnlversltAtsblbllothekRegensburg: Aufslellungssystematlk G: Germanl­ stlk, NladeriAndlsche Phllologle, Skandlnavlstlk. 3.Auf!. (Classification Scheme: German philology, Dutch and Scandinavian phl!ology, 3rd ad.) See also 2-261 (Orlg. de). Regensburg, DE: Unlv.Blbllothek 1987. 336p. 2-001 016 2-011 041-de-66 Beck, H.: Bookrevlewof Dahlberg,I.(Comp.& Ed.): International Classi­ UnlversltAtsblbllothek Regensburg: Aufstel/Jmgssystematlk P: Rechts· fication and Indexing Bibliography. Va1.3: Classification and Indexing wlssenschaH. 6.Autl. (Classilicalion Scheme P: Law. 6th ad.) (Orlg. de). Systems: theory, Structure, Methodology, 1950-1982. Indeks Verfag 1985. Regensburg, DE: Unlv.Blbtlothek 1987, 169p. XII,21 1p. (arlg.de). ZbLBibl.wes. Vo1.104, No.3, 1990. p.132-133 2-012 041·de-71/2 Unlversltatsblbllothek Regensburg: Aufstellungssystematlk Q: Wh1- 02 Literature Reviews schaHswlssenschaHen. 3,Autl. (ClassIficationScheme Q: Economics, 3rd ad.) (Orlg. de). Regensburg, DE: Unlv.Blbllothek 1987. 205p. 2-002 026 Biswas, S,C.: Smith, F.: Classod thesauri In Indexing and retrieval. A 2-013 041-de-98,-99 literature review and criticalevaluation of onUne alphabet/cclassaurus.(Orlg. Unlversltatblbllothek Regensburg: Aufstellungssystematlk B: Theologle en). Ubr.& Inlorm,ScI.Aes. VoLl1, No.2, 1969. p.109-1 41, 86refs. und Rellglonswlasenschan, 4.Aufl. (Classification Scheme B: Religion Following thepublication of the FIDleR Report "Compl.!terlzed Deep Struc­ and Theology. 4th ad.) (Orlg. de), Regensburg, DE: Unlv.Bibliothek 1987. ture Indexing Systems" (Oevadason, 1986) a research project entllied 287p, �Applicability and efficiency of Deep Structure Based Subject Indexing Languages (SILS): PRECIS vS.DSIS" was undertaken at DillS, Loughbo­ 2-014 041-de-94 rough University,-UK. The paper pmsents some of the drawbacks of the Unlversltatsblbllothek Regensburg: Aufstellungssystematlk LH-LO: Classaurus methodologywhich surfaced during the study, followad by some Kunstgeschlchte. 3.Autl, (ClassificationScheme LH-LO: History of Art, 3rd suggestions for fU!Ure lmprovement of the system. (Author, abbr.) ad.) (Orlg. de). Regensburg, DE: Unlv.Blbllothek 1988. 432p.

2·015 041-de-93 03 Dictionaries, Terminologies Unlversltatsblbllothek Regensburg: Aufstellungssyslematlk LD, LP-LX: Muslkwlssenschan, 3.Auf!. (Classification Scheme LD, LP-lX: Theory of 2·003 032 Music, 3rd ad.) (Orlg. de). Regensburg, DE: Unlv.Blbliothek 1988. 227p. Bles, W.: Book review of Sauppe, E.: WOrterbuch des Blbllotheks· 041-de-26/27, 522 wesen_ ... de-en, erHIe. MOnche": Saur 1988. XX.428p. (Orlg. de, en). 2-016 Lebende Sprachen. Vo1.3S, No.1, 1990.p.40-41 Unlversltatsblbliothek Regensburg: Aufstellungssystematlk V: Chemle und Pharmazle, 5.Aufl. (Class1flcaUon Scheme V: Chemistry and pharma­ 2-004 032 ceutics.) (Orlg. de). Regensburg , DE: Unlv.Bibllothek 1988. 112p. Burger, E.: Dictionary of Information Science. (Orig. en,de,fr, ru). 2·017 041-de-S1/4 Amsterdam,NL: Elsevier ScI. 1989. 903p. 2 vols. ISBN 0-444-98904-6. UnlversltAtsblbllothek Regensburg: Aufstellungssystematlk XA-YZ: Compilation 01 approx.2S,000 entries in English, German, French and und kllnlsche Madlzln, 4,Auf!. (Classification Scheme XA­ Russian, The following fields are covered:General foundations 01 Information TheoreUsche science, software development, hardware, system reliability, and applica­ YZ: Theoretical and dlnlcal medldne. 4th ad.) (Orlg. de). Regensburg, DE: tions of Information science In Industry and economics, VoU contains the Unlv .Blbllothek 1988. 197p. terms in the four languages with their codes,Vol.2has the indexes which refer to the codes. 2-018 041-de-39 UnlverslUitsblbllothek Regensburg: Aufslellungssystematlk R: Geogra· phle. 6.Aufl, (Classification Scheme R: Geography. 6th ad.) (Orlg. de). 2-00S 032 . Schulze, Hans H.: Computer-E.nzykloplidle - Lexlkon und Fachw6rter­ Regensburg, DE: Unlv.Blbliothek 1986. SOp. buch fOr Dalenverarbeltung undTelekommunlkatlon, 4 Bda. (Computer Encyclopedia - a lexicon and dictionary for dataprocesslng and telecommu­ 2·019 041-de-41 nication). 4 vols. (Orlg. de, en), Reinbek bHamburg: Rowohlt 1989. ca UnlverslUitsbJbllothekRegensburg: AufstellungssystemaUk W: Blologle 2900p. und vorkllnlsche Medlzln. 3.Aufl. (ClassificationScheme W: Biology and Contains some 26,000 entries with their English equivalents. preclinical medicine. :lrd ed.) (Orlg. de). Regensburg, DE: Unlv.Blbllothek 1968. 222p. 2-006 036 Ferre, A.-G.: Tarmlnologla de l'lnformatique. A data processingtermino­ 2-020 041-de-46,-47,-49 logy. (Orlg. fr, en), Meta. Vo1.33, No.3, 1986. p.427-438 UnlversltAtsblbllothek Regensburg: Aufstellungssyslematlk ZA-ZE: Land· und ForstwlrtschaH, Gartanbau. FlscherelwlrtschaH, Hauswlrt­ schaH. 3.Auf!. (Classification Scheme ZA-ZE: Agriculture, forestry, garde­ nIng, fishing, home economics. 3rd ed,) (Orlg. de). Regensburg, DE: 04 Classification Systems and Thesauri Unlv.BlblJothek 1986. 67p.

041 General Classification Systems and Thesauri 043 Dewey Decimal Classsilication 2-007 041, 645 lang, R. et al.(Eds): Canadian Thesaurus 1988: A guide to the subject 2-021 043.1 headIngs used In the Canadian Periodical Index and CPI Online. Trickey, K.: Book review of Comaroml. J, el al: Dewey Decimal Classi­ TMsaurus Canadien 1986. (Orlg. en, fr) . Toronto, Ont.: Info Globe 1988. fication and Relative Index, devlsad by Melvll Dewey, 20th ad. Albany, 465p. ISBN 0-92192S-04-2. NY: Forest Press 1989. (Orlg. en), cat.& Index., No.95, 1990.p.7-6 The bilingual list 01 over 35,000 terms covering all major subject areas and emphasizing Canadian topiCS reflects the contentsof magazines Indexed In 2-022 043.1 the Canadian Periodical Index. Henderson, KL.: Book review al Comamroml, J.el al.(Eds.): Dewey Decimal Classification andRelative Index. Devised by Melvll Dewey. 2.006 041 Ed,2O. Albany, NY: Forest Press 1 989. 4 vols. (Orlg.en). Cat.& Classll.Quart. Woods, W.E.: Manual and list of subject headlnga used on the Woods 10(1990)No.4, p.1 05-1 08 Cross Referenco cards. 6th ed. (Orlg. en). Evergreen Park, Il: Woods Ubr.Pub!, 1987. 132p. ISBN 0-912304-09-X. 2-023 043.1 Sweeney, R.: Dewey Decimal Classification · Edition 20. (Orig. en). 2-009 041-de-56 Outlook on Research Ubrarles. Vo!.11 , No,6, 1989. p.4-9 Unlverslt1itsblbllothek Regensburg: Aufstellungssystemallk D: P4dago-

In!. Classif. 17(1990)No.2 - Classification Literature 111 44 Library of Congress Classification locales contemporalnes. 2nd ed.Thesaurus W: standard ized vocabulary for the description and Indexing 01 oontemporary local administrative archi­ ves. (Orlg. fr). Paris: Dlr.Arch.France 1989. 249p. ISBN 2-86000-165-4. 2·024 044.6-667 Derham, L.O.(Comp.): Library of Congress Classification Class KDZ, 2-037 048·918 KG-KH: Law 01 the Americas, Latin America and the West Indies. Verdooth, A.: The written languages01 the world: A sUNey of the degree Cumulative Schedule and Index. (Orlg. en). Littleton, CO: Rothman 1988. and modes of use. Vol.3: Europa Occldentale!Western Europe. (Orig. 6642p. ISBN 0-8377-0127-9. en, Ir). Sainte·Foy, Canada: Presses de l'Universite Laval 1990. 672p. ISBN 2-7637-7222-6. 2·025 044.6-921 Burgess, M.: My".ry anddectectlv8 fiction In the libraryof Congress Claulflcatlon Scheme.(Orig. en). San Bernardino,CA; Borgo Press 1987. 2-038 048-984 184p. ISBN 0-89370-618-6. Elazar, D.H.: Elazar, D.J.: A classification system for libraries of Judalca. 2nd (Orig. en). Jerusalemllanham,MD: University Press of America 2-026 044.6-921 ad. 1968. 214p. ISBN 0·8191 -6583-2. Burgess, M.: A guide to science ,fiction and fantasy In the LIbrary of Congress Claaalflcation Scheme.2nd ad. (Orlg. en). San Bernardino, CA: Bargo Press 1988. 166p. ISBN 0-89370-827-5. 05 Periodicals and Serials

0448 Library of Congress Subject Headings 2-039 056.757 Expert Systems lor Inlormatlon Management. (Orig. en). Vol. 1 , NO.1, 2·027 0448-519 london: Taylor Graham 1989. Dickstein, R.: MUls, VA, Waite, E.J.: Women In Le's lerms: A thesaurus The journal will provide a forum for the exchange of information relating to the of Library of CongressSubject Headings relating to women. (Orlg. en). theory and application of expertsystem technology to Information processing Phoenix, AZ.:Oryx Press 1988. 221p. ISBN 0-89774-444-6. and provision of the field of Information science, busIness, the natural and social sclenoes, and the arIs and humanities. Its soope Includes expert

2·028 0448-93 systems; IntelUgent front ends; expertise directories; credit sanctionIng; Bratcher, P.: Smith, Jennifer (Comps): Music subject headings. (Orlg. en). auditing; patent sUNelllance; resource allocation; stock control; natural lake Crystal, MN: Soldier Creek P 1988. 323p. ISBN 0-936996-31-S. language Interfaces; records management; Information retrieval; office auto­ mation; portfolio management; business Intelligence; legal and regulatory advisory systems; network management. 048- Special Classification Systems and Editor: Forbes Glbb. Publisher: Taylor Graham, SOO Chesham House, 10 Regent Street, london Wl R SFA, U.K. Inthe USA: Suite 1 87, 12021 Wilshire Thesauri Boulevard, los Angeles. The follO'Ning articles have some bearing on our field of Interest: 2-029 048-33,36 2-040 Qulrchmayr, G.: Representing legal knowledge In Information Shlmomura, R.H. (Ed): GeaRef thesaurus and guide to Indexing. (Orig. systems. pA3-SO en). Alexandria, VA: Amer.Geo!og.lnst. 1989. 744p. ISBN 0-913312-98-3. 2-041 Zarr!, G.P.: Knowledge acquisition lor large knowladge basas using natural language analysis technique. p.8S- 109 2-030 048-S19 2-042 Black, W .J.:Johnson, F .C.: A practical evaluation oltwo rule-based leduc, S.: R4allution d'un thesaurus multilingue espagnol�franeal8 automatic abstracting tachnlquas. p.159-177 aur Ie Oro"s des femmes. Establishment of a multilingual thesaurus on 2-043 Neale, I.M.: MorriS, A.: Knowledge acquisition lor expert systems: Female rights. (Orig. fr, es). Paris: Inst.Natl.des Techniques delaDocumen­ a brief review. p.178-192 tatlon 1989. 63p. 2-044 Yannakoudakls, E.J.: Ayres, F.H., Hugill, J.AW.: An expertsystem lor quality control In cataloguing and dacumant Identification. p.120- 2-031 048-519 139 Szpakowska,J .-K.: Jeunes gens: thesaurus. Essal de systematisation des connaissances spedfiques de la classe d'age 13-25 ans. (Young people 2-045 057 thesaurus. Attempt at systematizing the special knowledge of the 13-2S UNITARIO. (Oflg. de). YoU, No.1, Bensheim, DE: Unltarlo Press 1990. years old). (Orlg. Ir). Montreal, Que.: Unlv.de Montreal 1968. 679p. In order to promote thenew universal language UNITAAIO, this first number gives an Idea of what It Is all about: Unltarlo Is �a synthesIs of three Ideas: 1) 2-032 048-538 the terminology key 01Eugen WQster, 2) the universal language Interllngua Bertrand-Baschwltl" VandeNeken, M.: Thesaurus specialise dana Ie (based on Italian), and 3) the Esperanto of ludwig lazarus Samenhor. traHement de la documentation des lnlections HIV et du SIOA. (Thesau­ Unltarlo Is meant as a universal language developed only lor use In scientific rus for documentation of literature on HIV Infections and AIDS.) (Orlg. Ir). oommunlcation. There is a European Center for the Promotion 01 Unltarlo at Bruxelles, BE 1968. 9Op. D-6140 Benshelm, P.O.B.1825. An Introductory text is available lhere also: (Mario Pleyer: Unltarlo - Entwurf elner Universalsprache. Benshelm: Unltarlo 2-033 048-56 Press 1989. 268p., DM44.-, ISBN 3-9802167-1-3). Unltarlo's grammar Is Mlnlstere de l'Education National, Centre National de Documentation available lor DM 2.- from the address above. P�agoglque: Thiaaurus MOTBIS. 3 Yols. (Orlg. Ir). Paris: CNDP 1990. 868p. The hierarchicallyorganized thesaurus comprises 10,781 entries olwhlch Its 06 Conference Reports. Proceedings 9,270 descriptors are arranged In 156 semantic lields. Seealso 2-280 2-034 048-822 Swan HIII,J.: Book revlewol: BlndlngTerms:A thesaurusfor use In rare 2-046 06.88-03-17/19 bookendspecial collections cataloguing. Chicago 1988 and Provenan­ Beck, H.: Book review of Wllle,R. (Ed.: KlasslflkaUon und Ordnung. Ta­ ce Evidence: Thesaurusfor use In rare bookand specIal oolJectJons catalo­ gungsband 12.Jahrestagg.d.Ges.f.Klasslf.eV, Darmstadt, 17.-19.3.1988. gIng. Chicago: Amer.lIbr.Assoc. 1988. (Orig. en). lIbr.& Inform.ScLAnnuaL Frankfurt Indeks Verl.1989. XII,396p. '" Stud.z.Klassll.Bd.19. (Orlg. de). VoLS 1989. 90-91 ZbLBibLwes. Vol.l 04, No.5, 1990.p.231-235 The two thesauri arereviewed together.

2-047 06.68-0S-13 048-823 Vlckers,J.A.: Book revlawof: Weinberg, B.H.(Ed.): Indexing: thastateof James, St.: Book review al: Peniston, S., Graham, T.: Thesaurus 01 our knowledge and the state 01 our Ignorance.Proc.2Oth Annual Meeting Inlormation technology terms. london 1909. 410 p. (Oflg. en). library Amer.Soc.lndexers,New York City, May 13, 1968. Medford, NJ: learned In­ Review. Vol.38, No.4, 1989. p.60 lorm.1989. X, 13p. (Orig. en). Indexer. Vol.17, No.1, p.71

2-036 048-824 2-048 06.88-05-13 Direction des Archives de France: Thesaurus W: Vocabulalre8 normall­ Fugmann, R.: Book review ofWalnberg, B.H.(Ed.): Indexing �The State sGs pour la deacrlptlon at l'lndexation des archives admlnlstratlves of the art and the state 01 our Ignorance. Medford, NJ: Learned Inlorm.19S9. 134p. «()fIg. en). Int.Classlf. VoL17, No.1, 1990. p.37·40

112 Int. Classif. 17(1990)No.2 - Classification Literature 2-049 06.88-06-28130 val systems. p.193-229 Rust,A.: Wissenschaftilche Konlerenz Fachtexle, Oeskrlptlonsmodelle 2-071 Brzozowski, J.P.: MASQUERADE: searching the full text of ab­ undsprachllcha Mlttal, Brandenburg, OCR, 2B.-30.JunI 1988. (Sdentlflc stracts using automatic Indexing. p.230-236 conference on special subject texts, description models and verbal tools, 2-072 Doszkocs, T.E.: CfTe MLM: Natural-language searching In an Brandenburg, GOR, 28-30 June 1988). (Orlg. de). Special Language - online catalog. p.263-279. Fachsprache. Vo1.11, No.1-2, 1989. p.71-73 Seealso below under 2-126 and 2·151

2-050 06.88-09-12115; 651/4 2-073 08.751 Scherrer,J.A. etal: Computerized natural medlcal languaga processing Smith, linda C.(Ed.): Questions and answers: Strategies for using the for knowledge ropr... ntatlon. Proc.IFIP-IMIA WG6 Intern. Working electronic reference collecllon. Papers presented at the 19B7 Clinic on Conf.on Computerized Natural MedicalLanguage Processingfor Knowledge library Applications of Data Processing, April 5-7. 1987. (Orlg. en). Urbana­ Representation, Geneva, 12-15 Sept.1988. (Orlg. en). Amsterdam, Nl: Champalgn,IL: Universityof illinois, Graduate of lIbr8J)'and Informa­ ' School Nordl-Holland 1989. 296p. ISBN 0-444-87356-2. tion Sdence 1989. 2OBp. The volume contains 15 papers lrom the �ClInlc� and from a preconference 2-051 06.88-11-1&'19; 43.06 session with an Introduction by linda Smith. Dack, D.: Australian attends conlerence on Dewey. (Orlg. en). Catalo­ 2-074 Creth, Sh.D.: Beyond technical Issues: The lmpa.ctof automation guing Australia. Vol.15, No.1, 1989. p.20-26, 1 ref. on library organizations. p.4·13 2-075 Hildreth, Ch.R.: Extending the access and reference service 2-052 06.69-10-214, 42.06 capabilities on the online public access catalog. p.14-33 UOC Seminar. (Orlg. en). lnt.Crasslf. Vo1.17, No.1, 1990.p.29 2-076 Roth, D.l.: Extending the online catalog. p.34-37 2-077 Baker, Betsy: Reference services and the networks: some reflec­ 2-053 06.89-10-516 tions on Integration. p.38-54 Heiner-Freillng, M.: Colloquy onthe DB Standard Subject File. (Orlg. en). 2-078 DIodato, V.: Online ready reference In academic libraries: current Int.Classlf. Vol.17, 1990, No.1 Int.Classlf.p.27-26. pracllces and a review of �annlng Issues. p.55-70 2-079Anderson, Ch.R.: Online ready referenceIn the public library.p.71· 2-054 06.69-12-18/21 84 Mater, E.: From Data Processing to KnowledgeManagement . Reporton 2-080 Rothenberg, D., Spencer, M.: Full-text databases: Implications for TopicalArea 1 of the 16th Intematlonal Colloquyat Erfurt, GDR, 1989. (Orlg. IIbrarl ... p.85-93 en). lnt.Classlf. Vo1.17, No.1, 1990.p.28-29 2-081 Noreault, T.: Optical publishing: Effects on reference services. p.94-102 2-055 06.90-06-14/17 2-082 Harter, St.P.: Onllna searching as ak problem-solving process. Tools for Knowledge Organization and the Human Interlace. Darm­ p.103-120 stadt, 14-17 Aug.1990.Program ofthaConfarence. (Orlg. en). lnt.Classlf. 2-063 Meadow, Ch.T.: Tailoring system design to users. p.121-131 Vol.17, No.1, 1990.p.31-33 2-084 Woodard, B.S.: Strategies for providing public service with an online catalog. p.132-143 2-085 Clark, Sh.: The online catalog: beyond a local reference tool. 07 Text Books (whole field) p.144-152 2-086 Watson, P.O.: University of illinois Library Catalog on CD·ROM: p.153-176 2-056 07.3 Impact on resource sharing and reference work. 2-087Clark, B.M., Bingham, K.H.:The new CD-ROM technology: shaping Beck, H.: Book review of Buchanan, B.: Blbllothekarische Kla8slflka· thefuture of reference and Information research. p.177-187 tlonasthoorl•. Obers.{aus dem Engl.) von U.Relmer-BOhner. Munchen, 2-088 Mlscho, W.H.: etc.: Saur 1989. 151 p. (Orig. de). ZbJ.Blbl.wes. Vol.l04, No.5, 1990. p.235- An end user search service with customized •. p.188-204 237 Interface softwar

2·057 07.3 Cochrane, P.A.: Book review of Hunter, E.J.: Classillcation made sim­ 09 Standards, Guidelines ple. Brookfield, VI.: Gower Publ.1986. .11Sp . (Orig. en). Llbr.& Inform.Sd.Annual. Vol.5, 1989. p.87 2-089 09.853 British Standard Recommendation for Preparing Indexes to Books, Pe­ 2-058 07.3 riodicals, and other Documents. (Orlg. en). London : Brit.Stand.tnst. 1988. Laresse, E.: Guide du classement dans l'entreprlsa.(Classlflcation guide 17p. BS 3700:1988. In theenterprise.) (Orlg. fr). Paris: Ed.d'Organlsal. 1988. 173p. ISBN 2-7081- 0952-9. 2-090 09.268 US MARC Code List for Languages. (Orlg. en). Washington, DC: lIbr.Congr.Gataloglng DistrIbution SaNlce 1989. 08 Other Monographs This list was designed to beused wIth the US MARC Format for Bibliographic Data. 2-059 08.75 Willen, P. (Ed.): Document retrieval tems.sys (Q-Ig. en). London: Taylor Graham 1988. 292p. ISBN 0-947568-21-2. 1 Theoretical Foundations The volume contains 16 readIngs - partly of historic Interest - among which the following: 2·060Cleverdon, C.: Optimizingconvenient online access to bibliogra­ 12 Conceptology in Classification phic databases. p.32-41 2-061 Salton, G.: Automatic Indexing and abstracting. p.42-80 2-091 121; 757 2-062 Harter, St.P.: Statistical approaches to automatic Indexing. p.81- Bestougeff, H., lIgozat, G.: Plausible reasoning In expert systems. (Orlg. 98 en). In: ExpertSysl.: Theory and AppJ.: lASTED InI.Conf., Geneva, 16-18 2-063 Lennon, M., Peirce, D.S., Tarry, B.D., Willett, P.: An evaluation of June 1987: Proc. Anaheim, CA 1987. p.116-219, 8 refs. some contlatlon algorithms for Information retrieval. p.99-l05 2-064 Walker, SI.: Improving subfect access painlessly: recent work on 124 Knowledge Representation the OKAPI online catalogue projects. p.113-123 2-065 Sparck Jones, K.: A statistical Interpretation of term specificity Seealso 2-040 and It_ appUcatlon In retrieval. p.132-142 2-066Croft, W.B.: Harper, D.J.: Using probabilistic modelsof document 2-092 124 retrieval wHhout relevance Information. p.161-171 Aguirre, J.l.: A taxonomic contructlon system based on an obJect­ 2-067 Bersteln, l., Williamson, R.E.: Testing of 8 natural language retr/a­ centered knowledge representation. (Orlg. en). In: Expert Sysl.: Theory val system for a fuU toxt knowledge base. p.25O-292 and Appl.: lASTED Int.Conf., GENEVA, 16-16 June 1987: Proc .. Anahelm, 2·068 Frakes,W.B.: Term conllatlon for Information retrieval. p.106-1 12 CA 1987. p.22-224, 13 fefs. 2-069 Porter, M., Galpin, V.: Relevance f&8dback II} a public access catalogue for a research library: MUSCAT at the Scott Polar Institute. 2-093 124 p.172-191 Font, J.-M.: les graphesde concept.(The graphs of the concept.) (Orlg. fr). 2-070 Salton, G., Pt.4cG1I1,M. J .: The SMART and SIRE experimental retrle- BulLdu CID., No.34-35, 1989. p.1 11-116

In1. Classif. l1(l99Q)No.2 - Classification Literature 113 2-094 128-821 Reply to a discussion published by A.BOhler and the author In Conceptus Marloth, H.: Book review of Folberth, O.G., Hackl, C.: Dar Informations­ 23(1989)No,59, p.91-saand 99-102. begrlff In Technlk und Wissenschaft. MOnchen-Wlen: Oldenbourg 1986. 299p. (The Information conceptIn technology and sdence. (Orlg. de). 2-10S 164 Mltt.bI.BlbI.NRW. Vol.39, No.3, 1989. p.374-376 Bunlyatov, A.A.: Onthe correlallon 0' thehistorical and the logical I" the development 0' science. (Orlg. ru). Baku: ELM 1988. 141p. 13 Mathematics in Classification 17 Problems in Classification 2-095 132 See also 2-140 Baumann, E.V.: Fuzzy classification methods (a variational approach). (Orig.en).Automation & Remote Control. Vo1.49, No.12, 2, 1989. p.1650-61 , 2-107 177 13 refs. Salomon, K.: The Impact of CD-ROM on reference departments. (Orlg. en). RO. Vo1.28, NO.2, 1988. p.203-215 Results of a nationwide survey to refer�nce chairpersons In 150 US acade­ 15 Psychology and Classification mic libraries.

2·108 177 2·096 151 Woodward, T.: Information retrieval uslng.mlcros. (Orlg. en). Asllb Proc. Benedlktsson, D.: Hermeneutics: Dimensions toward LIS thinking. (Orlg. Vol.41 , NoA, 1989. p.157-162 en). lIbr,& Inform ,ScI.Res. Vo!.11, No,3, 1989. p.201·234, 33refs. Among the �exdtlng" developments of microcomputers - considering the In­ 156 creasing power and flexlbll!ty of their IR packages · are also their use of Knowledge Acquisition authority files and thesauri, browsabilityof Indexes, more flexible outputs and See also 2-041, 2'043 their Integration with RDBMS and SOL. Prospects In particular towards hypertext systems are outlined. 2-097 156 01 Manzot, M., Rlcd, F., Batlstonl, A., Ferrari, C., Trucco, E.: Embedding functional knowledge In an actor network. (Orlg. en). In: Expert Syst.: Theory and Appl.: lASTED Int.Conf., Geneva, 16-18 June 1987: Proc. 19 History of Classification Anaheim, CA 1987. p.167·169, 7 refs. 2·10Ba 191

2-098 156 Innocenti, p,: Notes concerning thehistory of dasslflcatlon. (Orlg.!I). L'lndl­ Karbach, W.: Modellbaslerte Wissensacqulsitlon. (Model-based know­ clzzazlone 4(1989)No,1, pA7-63, 17 refs. ledge acquisition). (Orlg. de) KOnsll.lntelUgenz. No.4, 1989. p.13 Considers the results of the conferenceheld In Ottawa In 1971 In the light of Explication of the concept cited In the title. the search for a synthesis of philosophical prlndples with those relevant to library science. Offers and evaluates several examples of empirical classifi­ cation taken from andent catalogues. 2-099 156 (Author, abbr.) Laske, 0.: Ungel�8te Probleme bel der Wissensacqulsillon fOrwlssens­ 2-109 194; 485 baslerte Systeme. (Unsolved problems of knowledge acquisition for know· ledge-based systems). (Orlg. de). KQnsll.lntelilgenz., No.4, 1989. pA-12, 11 Ochmanska, A.: Changes In O.Hartwlg'sclasslflcallon In the university (Orlg. pi). Blbliotekarz. refs, libraries 01 Halle, GDR, and Poznar'!, Poland. Three perspectives of the establishment of knowledge-based systems are Vo1.55, No.12, 1988. p.3-6, 12 refs. presented, starting out with rapid prototyping to reach the maintaining of a knowledge management pointof view. Difficulties In knowledge acquisition 2- 110 198-825, 43 for KBS are examined In the light of epistemology. Next to purely technical Lamonde, Y.: A universal classification for the study of nineteenth solutions, such a hypermedia systems, management solutlons are discus­ century libraries and booksellers. (Orlg. en). LIbraries & Culture. Vol.24, sed, which presuppose a rethinking of KBS. No.2, 1989. p.I58-197, 17 refs.

2·100 156 Legrand, J.: Contribution to the achievement 0' a syntax In knowledge 2 Structure and Construction of engineering. (Orlg. en). In: Expert Sysl.: Theory and Appl.: lASTED Int.Conf., Geneva, June 16-18, 1987. Anaheim, CA 1987. p.45-49, 33 refs. CS and T

2-101 156 Shaw, M.L.G., Gaines, B.A.: Comparingconceptual structures: consen­ 21 General Questions of CS and T sus, conflict, correspondenceand co ntrast. (Orlg. en). KnowledgeAcqul· sltlon. Vol.1 , 1989. p.341-363, 21 refs. Seealso 2-002

2-102 156 2-1 11 214; 761 Soda, G.: A flexible editor for knowledge acquisition. (Orlg. en). In: Buchan, R.L.: IntertWining thesauri and dictionaries. (Orig. en). ExpertSyst.: Theory and Appl.: lASTED Int.Conl., Geneva, l S-1 8June 1 987: Inform.Servlces & Use. Vo1.9, No.3, 1989. p.171-17S Proc.. Anahelm, CA: 1987. p.135-139, 8 refs.

2·103 156 22 Elements of CS and T Comtesse, X.: Bader, A.: Un modele d'acqulsltlon connalssances base aur re role organluteur des schemes. (A knowledge acquisition model See also 2·218 based on the organising role of schemes). (Orlg. fr) . ln: Expert Syst.:Theory and Appl.: lASTED Int.Conf., Geneva, 16-18 June 1987: Proc.Anahelm, CA 2-112 224 1987. p.140·143, 9 refs , Bara, I.: Problems of modelling and terminological control In documen­ tary Information retrieval systems. A reexamlnatlon of the thesaurus concept. (Orlg. en). Probl.lnform.sl Doc, Vol.23, No.3, 1989. p.l01-105, 6 16 Development of Science and Classification refs .

2·104 161 2-113 225; 745 II'ln, V.V.: (Orig. (u). Moskva: Vyssh.shk. Scientific knowledge criteria. Iyer, H.: Natural language representallon: Transformallonal rules. (Orig. 1989? 128 p. en). Int.Classlf.VoI.17, NO.1, 1990. p.9-13,9 refs.

1627 2-105 2-114 225: 6761 Jakowlewitsch, D.: O ber elne unzulAngllche Auffas8ung dar methodolo­ Horvathn6, Z.: A faceted classlflcallon system In the library of the glschen ElnheH dar Wissenschan und Ihr. unbefrlodlgenda Verteldl· PolytechniC 'or Indust,lal Design. (Orig. hu). KOnyvtari Flgyelo. Vo1.35, gung. (On an Inadequate Interpretation of the methodological unity of No.1, 1989. p.52-64 science and Its dissatisfying defense). (Orlg.de). Conceptus. Vo1.24, No.61 , 1990. p.1-103

114 Int. Classif. 17(1990)No.2 · Classification Literature 2-1 15 226 2-128 247 Risko, A.: Linguistic Interpretation of sUbject headings. (Orlg . sk). Knlz.a Gamaf-Eldln, M.S.: Gomer, Th., Ramez, E.: Integrating relational databa- ved. Inform. VO!'21 , No.3, 1989. p.99-103, 6 refs. 88S with support for updates. (Orig. en). in: Inl. Symp. Database and Distribution System, AusUn, TX, Dec.5-7, 1988: Proc. Washington, DC 1988. p.202-209, 20 refs. 23 Construction of CS and T

2-116 234 25 Numerical Taxonomy Mamede, H.C.A., sergio, D.F.: Relevant aspects In designing languages to aCC88S bibliographic databases.(Orig. pt). Cienc.!nform .. VoL17, NO.2, Seealso 2-123 1988. p.1 19-121, 7 refs. 2-129 253 2-117 235 Broder, A.J.: Strategies for efflclent Incremental nearest neighbor Blellcka, l.A.: Thesaurusconstrucl lon:method and organisation. (Orlg. search. (Clflg. en). Pattern Recognition. Vol.23 , NO.1/2, 1990. p.171-178, 7 pi). Cent.lnlorm.nauk.tecl1nJ ekon. No.67, 1988. p.1-69 refs .

• "6 _ 2-130 253 SvenonJus, E.: Design of controlled vocabularies In the context of Huang, J.S., Shieh, W.R.: A heuristic method for separating clusters emergIng technologies. (Orlg. en). Libr.Scl.SlantDoc. Vo1.25, No.4, 1988. form noisy background. (Orlg. en). Pattern Recognition Vo1.23, NO.1I2, p.21S-227, 12 refs. 1990. p.147-157, 12 refs.

2-119 236 1-131 253 Rees-Potter, l.K.: Dynamic thesaural systems: a blbIlometrlc study of Shlaw-Dong, D.l., Casasent, D.P.: iterative Fisher/Minimum-Variance termInological and conceptual change In sociology and economics. optical classlller. (Orlg. en). Pattern Recognition Vo1.23, NO.314, 1990. with application to the design of dynamic thesaural systems. Information p.385-391, 21 refs. Processing and Management 25(1989)No.6, p.677-692, 29 refs. The study Investigates a mechanism by which thesauri can be updated and 2-132 255 maintained using citation analysiS, co-cllation analysis and dtallon context Lenz, R.: Qroup Invariant pattern recognillon. (Orlg. en). Pattern Recog­ analYSiS. Data ln twenty-six speciallyareas In economics and sociologywere nllion. Vol.23, NO.1/2, 1990.p.1 99-217, 22 refs. used. Experts in these disciplines conclude that the terms used were appropriate for describing the specially areas. (Author) 2-133 255 Carman, C.S., Merlckel, M.B.: Supervising lsodata with an Information 2-120 237; 44.8 theorellc stopping rule. (Orlg. en). Pattern Recognition. VoL23, No.lf2, Lazinger, S.S.: Producing an LCSH authority list for special libraries 1990.p.18 5-197, 17 refs. with dBase. (Orlg. en). Electronic libr. VoL8, No.1, 1990.p.8-14 2-134 255 2-121 237, 651/4 Pedrycz, W.: Fuzzy sets In pallern recognition: methodology and Bellamy, l.M.: Thesaurus development for SUbject cataloging. (Orig. methods. (Orlg. en):Pattern Recognition. Vol.23, No.1/2, 1990. p.121-146, en). Spec.libr. Vol.SO, No.1, 1989. p.9-15, 28 refs. 85 refs.

24 Relationships 26 Notation. Codes

�1� �1 2-135 265 Johansen, Th.: Methods of subject-structure display. (Orlg. en). Sallla, M.P., Agrawal, S.P.: Book numbers: some Indian methods. (Orig. Int.Classlf. Vol.17, No.1, 1990. p.2-7, 8 refs. en). New Deihl:Concept Publ. 1990.96p. Concepts In Communication N.6.

2-123 241; 723;255 2-136 267 Kozlova, I.V.: A probabilistic-statistical model of Index terms' semantic lIu, Songqlao: Online classification notation: Proposal for a Flexible relations In a problem-oriented database. (Orlg. en). Faceted Notation System (FFNS). (Orlg. en). Int.Classif. Vo1.17, No.1, Avtomallz.lnform.bibl.protsessov v vyssh.shk. Moskva, SU 1989. p.41-46, 5 1990.p.1 4-2O, 16 refs . refs.

2-124 242; 753 28 Compatibility and Concordance between Lu, Xln: Document retrieval: a structuraf approach. (Orlg. en).Inform.Process.& Management.Vol.26, No.2, 1990. p.209-218, 18 refs. Indexing languages In the model described, documents and query statements are structurally coded In order to take Into account any hierarchy or ordering among the Seealso 2-202 conceptual coordinates and are structurally matched by using the algorithm that cannot be expressed In a form of equation. (Author, abbr.) 2-137 264 Kralova, A: Experiments In the use of the Balex databases of develo­ 2-125 243 ping concordances between Indexing languages. (Orlg. ez). Knlz.a Chowdhury, G.G.: Mahapatra, M.: Appllcatloons of the Theory of Relatio­ ved.lnform. Vol.21, No.3, 1989. p.104-108, 6 refs. nal Analysla In Inlormallon retrieval. (O(lg. em). J.Ubr.& Inform.ScI. Vo1.14, No.1, 1989. p.1-12, 23 refs. Compared to earlier application fields of Farradane's Theory of Relational 29 Evaluation of CS and T Analysis, five new areas of research on relational Indexing are suggested, I.e. compilationof adetalled manual, wldeappllcallon In scientific, social sdence, 2-138 296 and humanities subject literature, study of the nature of subject literature, Klshlda, K., Mushakoll, s., Malsuyama, N., Higuchi, K. etal: A comparative comparative effldency of relationat lndexlng, and oompllatlon of a relational evaluallon of thesauri: performance of 'presentation 01 a body of thesaurus. (Aulhors) knowledga'. (Orlg. la). Joho no Kagaku to GiJutsu. VoJ.38, NO.10, 1988. p.565-572, 14 refs. 2-126 246 Robertson, S.E., SparckJones, K.: Relevance weighting of search terms. (Orlg. en). DocumentRetrieval Systems.London, GB: Taylor Graham 1968. p.143-160, 21 refs. 3 Classing and Indexing

2-127 247 31 Theory of Classing and Indexing Coussoulet, J.-P.: Conception et niallsatlon d'un system. relatlonnel documentalre: Campus-Doc. (Conception and establishment of a relatio­ also2-047, -048,-056, -057, -058, -347 nal documentallon system: Campus-Doc). (Orlg. tr). Nice, FR: See Unlv.d.Sclences 1988. Th.3Jeme cycle: Informatique.

In!. Classif. 17(1990)No.2 - Classification Literature 115 2·139 311 2-152 342 Goncharuk, E.D., Kondrat'ev, LA.: Quallly conlrol ln Information rel rieval Blru, T., EI-Hamdouchl, A., Rees, R.S., WIllett, P.:lncluslonof relevance In­ Iyeteme: monitoring document Indexing. (Orlg. ru). Nauchno­ formation In the term discrimination model. (Ong. en). J.Doc. Vo1.45, takhn.1nform.,Ser.2.,No .11, 1986. p.17·19, 3 refs. No.2, 1989. p.85-109, 22 refs.

2·140 311.173 2-153 342 Ramalah, LS.: Indexing witch-hunt and the failure of language. (Orig. Nelson, M.J.: Stochastic models for the distribution of Index terms. en). IASUC Bull. Vo1.32, No.3, 1987. p.99-103, 13 refs. (Orlg. en). J.Doc.Vo1.45, No.3, 1989. p.227·237, 25 refs.

2-141 313 2-154 342 Bles, W.: Book revlewof Richter, Noe: Grammalre del'lndexation alpha­ SrInivasan, P.: A comparison of two-poisson, Inverse document Ir... b4Uque(grammar for alphabetical Indexing). Le Mans: BibHotMque de quency and discrimination value modelsof document representation. l'Unlvarslte du Maine 1985. 4th ed. 155p. (Orlg.em). Int.Classif. Vol. 17, NO.1 , (Orlg. en). Inform.Process.& Management. Vol.26, NO.2, 1990.p.269·278, 1990. p.40-42, 2 refs. 13 refs.

2·142 313 2-156 343 Llndtner, I.: Vor- und Nachtelle belmIndexleren mit Informatlonsrecher­ Jones, L.P., Bessonet, C.de, Sukhamay, K.: ALLOY: an amalgamation of cheaprachon vom Klassllikationatyp 1m Hlnbllck auf Ihre Anwendung expert, linguistic, and statistical Indexing methods. (Orlg. en). In: 11th In der ODR. (Advantages and disadvantages of indexing with classificallon InI.Conf.Res.& Dev.lnform.Retrleval, Grenoble, 13·15June 1988. Grenoble, systems with respeC1 to thalr application In the GDR). (Orlg. de). Berlin, DD: FR 1988. p.191-199, 13 refs. Humboldt-Unlv. 1988. 159p. Dissertation. Description of contents In Zbt.Blbl.wesen 104(1990)No.5, p.227 2·157 3435 Kimoto, H.: Nagata, M., Kawai, A.: Automatic Indexing system for Japa­ �Ia ��m nese text. (Orlg. en). Rev. Electr. Commun. Lab. (Japan) . Vo1.37, No. 1, Rapp, A.: Wattler, M.: Simulation der Suchwortgenerlerung 1m Informa­ 1989. p.51-56 tloln Retrieval durch Propaglerung In eloomkonnektlonlstlschen Wort­ netz. (Simulation of search term generation in information retrieval by 2·158 345 propagation In a connectionist lexical net). (Orig. de). Nachr.Dok. VoL41 , Fagan, J.L.: The effectiveness of a nonsynlactlc approach to automatic NO.1, 1990. p.27·32, 9refs. phrase Indexing for document retrieval. (Orlg. en). J.Amer.Soc.lnform.ScI. Vol.40, No.2, 1989. p.115·132, 43 refs.

32 Subject Analysis 2-159 345 Scott, SI.: Keywords and Dewey: the belt-and-braces approach to (Orlg. en). Audiovisual 2-144 324; 733 Information In Moray House AV Resource Centre. Llbrn. VoI.1S, NO.1, 1989. p.42-44 Schwarz, Ch.: Content basedtext handling. (Orig. en). Inform.Process.& Management.Vol.26, No.2, 1990.p.21 9-226, 10 refs. Research and development In thefield of AI techniquesapplicatlon to natural 2·160 346 language processing at theSiemens Corporate Research Laboratories are Panevov!i, M.: PASSAT Indexing language. (Orig. cs). Cs.inform. VoL31 , desCflbed. No.4, 1989. p.l02-105, 6 refs.

347; 751 2·145 326 2-161 Bates, M.J.: Rethinking subject cataloging In the online environment. Fedoseeva, T.V.: Applying classificatory techniques to the description (Orlg.en). Library Resources & Techn.SeN. Vo1.33, No.4, 1989. p.400-412, of an obJectographlc database subject domain. (Orig. ru). MezhotrasLinform.sluzhba., No.2, 1989. p.2D-22, 6 refs. 19 refs. The design of a "superthesaurus", as parto'a friendly front-end user Interface Is described. Such a thesaurus Is geared to the needs of users rather Ihan 2-146 326; 753 indexers and Incorporates the findings of recenl research on the patterns of MacLeod,LA.: Storage and retrieval of structured documents. (Orlg. en). subject desCflption by searchers. lis design also reflects the different Inform.Process.& Management. Vol.26, NO.2, 1990,p.1 97·208, 22 refs. demands of onllne searching as opposed to manual searching. (Author, abbr.) 2·147 326 Palce, C.D.: Constructing Uterature abstracts bycomputer: techniques 2-162 348 and prospects. (Orlg. en). Inform. Process. & Management. VO!.26, NO.1, Fox, EA, Nunn, GL, Lee Whay, C.: 1990.p.171-166, 52 refs. Coefficients for combining concept classes In a collection. (Orlg. en). In: 11th Int.Conf.Res.& Dev.lnform.Retrieval, Grenoble, 13-15 June 1988. Grenoble, FR 1988. 2·148 326 p.291-307, 37 refs. Takahlro, A.: Kazuteru, D., Yuujl, S. et al: A stUdy ol a retrieval system lor technical paper.using document structures. (Orlg.ja).Kogakuin daiga­ ku kenkyu hokoku, NO.66, 1989. p.216-220, 17 refs. 35 Manual and Automatic Ordering Techniques 2·149 328-66 Rhodes, S.: Understanding legal language: a pragmatlc approach. See also 2-192 (arlg. en). In: IPCC'88: Int.Prof.Commun.Conf., Seattle, WaSh., Oct.5-7, 1988: Conf.Rec.New York 1988. p.313·317, 11 rels. 2·163 351 Fidei, A.: Organizing, Indexing and retrieving Information. (arlg. en). School Ubr.MedlaQuarl. Vol.17, No.4, 1989. p.206-209 33 Classing and Indexing Techniques 356 HypennedJa 2-150 337; 752 Dykstra, M.: PRECIS In the online catalog. (Orlg.en). Cal.& Classlf.Quart. 2-164 356 VoLl0, No.1/2, 1989. p.81-94, 7 rels. Onodera, N.: Hypertext: computerized organlsallon of knowledge. (Orlg.ja). Joho kanrl. Vol.31, No.ll, 1989. p.961·977, 12refs.

34 Automatic Classing and Indexing 2-165 356, 752 Kinnell, S.K., Richards, T.: An online Interface wllhln a hypertext system: Project Jefferson's electronic notebook. (Orlg. en). Online VoL13, No.4, See also 2-061, -062, -065, -066,-068, ·070, -071, -336 1989. p.33-36 2·151 341 2·166 356 Perry ,SA, WIllett, P.: A review oftha use 01 Inverted fll8s for best match Plalsanl·Swenn, K.: L'lnterface utilisateur d'un systeme HypermedIa: I'exem­ searching (Orlg. em). In: Document In Information retrieval systems. pie d'Hypertles. (The user Interface of a hypermedia system: an example of Retrieval Systems. london: Taylor Graham 1968. p.124-131, 28 refs. Hypertles). (OrlgJr) Bull du CID 1989, No.34·35, p.99-104.

116 Int. Classif. 17(1990)No.2 - Classification Literature .,� � 2·182 388-99 Shlmamura, T.: Indexing and Illing order: access points to the titles of HOgemann, S.: Der Schlagwortlndex als Konkordanzreglsler zur SIB, Japanose periodical •. (Orlg. la) . Blblos (Japan). VaI.S8, No.1 1 , 1987. p.19- ASB und SSO. Elne Untersuchung am Beispiel des Fachgebieles Chrlstll­ 25, 4 refs. che Religion. (The subject Index as a concordance to three German dassl· ficatlon systems (SfB, ASB, SSD). A study In the field of Christian Religion). �oo � (Orlg. de). KOln, DE: Fachhochschule Blbllotheks-und Dokumentatlonswe· Weldenleld, G.: Quelquesexample. d'utlllsations d'hypertext8•. (Some sen 1988. Dlplomarbelt Ott.Blbl. examples of hypertext application). (Orlg. Ir). Bull.du CID. No,34�35, 1989. p.l0S-tOg 2·183 389 Bertrand·Gastaldy,S.: Evaluation de l'lndex 1984/85 du Journal des 357 Document Structuring debate de" A888mbIHnatlonaladuOuebec.(Orlg. Ir) . �ntreal, Canada: Unlv.de Montrllal l987. 151p. •,M m 2·184 389 Andr'. J.: Reconnalsaance de 1a structure loglque des documents. (Reoognlt1on of the document's logical structure). (arlg. Ir). Bull.dU CID., Kocere,W.Va., Kresllnsh, K.A., laure, I.G,: Currant bibliographIc Indexes No.34-35, 1989. p,47-52, 7 refs. of theLatvian Academy of Sciences main libraryand Investigation Inlo their eflectlvoness. (Orlg. ru). Inf.·blbl.obsluzh,uchenykh, tv1oskva, SU .,M m 1988. p.I50-163 Surreau, L: Structur. de.documents: standardODA . (The structure of documents: the OOA standard) (Office Document Architecture). (Orlg. Ir). 2·185 389 lNlttmann,C.: Bull.du CID, No.34-35, 1989. p.7-13 Subheadlnga In award�wlnnlng book Indexes: a quantlta· live evaluation. (Orlg. en), Indexer. Vol.17, NO.1, 1990.p.3·6, 8 refs. 2-171 357 Tazl, S.: Roconnalaaance de la structure taxtuelle des documents. (Recognition olthatextual structure of documents). (Ong. Ir). Bull. du CID., 39 Evaluation of Classing and Indexing No.34·35, 1989. p.37·45, 11 refs. Seealso 2·066, -067, ·247

2-186 391 38 Index Generation and Programmes Chu, C.M" Ajlferuke, I,: Qualily 01 Indexing In library and Informallon sclencedataba888, (Clflg, en), Online Rev. Vol.13, NO.1, 1989. p.11-35, 17 .,n m refs. LI, WeI: Design 01 Index to reference tools. (Orig. ch). Tushu Qlngbao Gonzuo., No.4, 1987. p.26·27 .,� �1 Giger, H.P.: Concept based retrieval In classlcal lR systems. (Orlg. en). 2·173 382; 388·51/4 In: 11thInf,Conf.Re s,& Dev.lnform.Retrieval, Grenoble, 13·15 June 1988. Van Llew,H.D., Gordlch,M.J., Wen, N.· T., Sung, P.-F.: UsIng MEDLINEand Grenoble, FR 1988. p.275-289 I personalcomputarto genaratea cumulative Index lor aJournal. (Orlg. en). Serials Review. Vol.15, NO.1, 1989. p.27·34 �oo �1 Nalh, P., SIngh, S.N.: Inlormatlon explosion and Indexing systems. 2-174 383 (o...lg. en), Progress Ubr.& Inf.ScI. Vo1.9 1989. p.1 9-25, 14 refs. MlttaJ,A.: Creation of 8ubJect indexto NUCSSI database with CDSIISIS Stresses the need for evaluatlnng Indexing systems. Describes briefly the package. (Orlg. en). Annals Llbr.Sci.& Doc. Vol.36, NO.1-2, 1989. p.52·62, development of indexing systems. Explains various types of Indexing sy- 3 refs. stems and highlights their future. (Author)

.,� = 2-189 391 Schmidt, J.M., WIlson, J.S.: Oul of the shoebox: publishing a regional Pao, M.L, Worlhen, D.B,: Retrieval offectlveness by semantic and cUa­ perfodlcal lndex using I microcomputer. (Clflg. en). Reference Services tlon .earchlng, (Otlg. en). J.Amer.Soc,lnform,Sd. Vol.40, No.4, 1989. Review. VoI.16, No.4, 1988. p.13-20, 7 rels. p.22&235, 32 rels,

2.176 383 .'00 = Shulchl, U.: Index construction, Indexing and the thesaurus. (Orlg. ja) . Chan, LM.: Inler·lndexerconsistency In subject calaloglng . (Orlg. en). Joho-no kagaku 10 gijutsu. Vol.38, No.1 1, 1988. p.637-643, 8 refs. Inform.Techno1.& Llbf" Vo1.7, No.2, 1989. p,349-358, 8 refs.

2-177 386 .,9, = Ngah, lalnab Awang: The Malayslana Conference Collection on CDSI Maruyama,Sh.: Metrics 8. I. carte 2: consistency rate of classilicallon, ISIS: the second ph&80. (Clflg. en). Kekal Abadl. Vo!.7, No.4, 1988. p.6-15 (Orlg.la). Joho no Kagaku to Gljutsu. Vo1.37, No.5, 1987, p,198·199, 5 refs,

2·178 387 2·192 394; 354 lNlHlams, D.E.: Concept Finder. (Orlg. en). Techn.Servlces Quart, Vo1 .6, Maruyama, Sh.: Metrics 8.la carte 3: the elllclency 01 shelf classlflcaUon No.2, 1988, p,74-77 and Its llmltallons. (Otlg.ja).Johono KagakutoGIJutsu. Vol.37, No,6, 1987. p.243-244 .,� - Meiss,S,: Thesauru., lntegrlart. onlineund dynamlsch. Neues Arbelten 2-193 396; 842 mit DOMESTIC, (Thesaurus, Integrated, online and dynamical. A new kind of Krause, J., Wamser-Hacker, Ch.: PADOK·II: Retrlevallesls zur Dewar· work wllh Ihe DOMESTIC software). (o...lg, de), MOnchen, DE: KTS lung von Voillextindexierungsvarianien lOr da. Deutsche Patentlnfor· Inform.Syst. 1989, 103p, ISBN 3-9801407-3-3, matlonssystem. (Retrieval test for the evaluation of fulltext Indexing variants of the German Patent Information System). (Orig. de), Nachr.Dak .. Vol.41 , .,� - No,1, 1990. p.13-19, 28 refs. Ritzier, C.: Verglelchende Untersuchung von PC-The88urusprogram· men. (Comparative studies of PC thesaurus programs). (Orlg. de). Darm­ .,M - stadt, DE: Fachhochschule. Fachberelch Information und Dokumentatlon Rlesthuls, G., Stuurman, P.: Tendenzen In de oderwerpsontslultlng: (2) 1989, 200p., 151 refs, vrfJe termlnologle versus gecontroleerd vocabulalre. (Trends In subject Indexing: (2) free terminology versus a controlled Indexing language), (Orlg, 2.181 388-68 nl). Open Voi.21 , NO.l0, 1989. p.348-352 Bourret, A., Poole, E.: Index sur Is gue"e .t 'a palx: Index relatlonnel de motachia pourSONJA. Banque dedonn�s lnlormatlsees sur la guerre, la .,� - palx et Ie dllsarmement en langue francalse, (Index on war and peace: a Schwandl, J,: Nulzung konirolllerterund fr.'erTermlnologle bel derRa­ relational Index of keywords for SONJA, a database with Information on chercM In grosSGn blbllographlschen Oatenbasen. (Util1zatlan of con­ peace 8I1d disarmamentin the French language). (Orlg, Ir), OUllbec, Catla­ trolled and free terms In searching large blbllographfc databases) , (Orlg, de). de.: Unlv.Laval l989. 26p,"" Gah.lechn,du Groupe de Recherche surla palx. Informatik. Voi.36, No.4, 1989. p,136-138, 12 refs. ISBN 2·9801653-{)-1.

Int. Classif. l7(1990)No.2 - Classification Literature 117 4 2-209 44.6-66 On Universal Systems Bennett. A.A.: The new echeduleafor Law and th. Implications lor clas­ sification. (Orlg. es). lnv9sllgaci6n Blbllotecol6grca. Vol. 1 , No.3, 1967. p.3- 42 On the Universal Decimal Classification 20. 17 refs. See also 2·052 2-210 44.6-007 Aderlblghe,M.R.. Udoh, D.J.E.: L.C.Subcla.. PL8000-8844: A CIS. for 2-196 42.07 I'fIvl.'on .(Orlg.en).Cat.& ClassltQuart. Vol.10,No.3, 1 990.p.77-90, 6 refs. Anderson, J.D.: Book review of BOS8, H.: Universal Decimal Cla.slllca· tlon: Theory and practice. New Deihl: Sterling 1987. 176p. (Orlg. en). lIbr.& Inform.Sd.Annual. Vol.5 1989. p.85 448 Library of Congress Subject Headings 2·197 42.1 Mishra, O.S.: UDC In perspecllve. (Orlg. en). Progress lIbr.& Inform.ScI. See also 2·120 Vo1 .9. 1989. p.19-25++, 42 refs. Discussesthe purposeand origin of growth of the UDC. Describes the basic 2·21 1 446.3 structure of the UDC and also Its notation. Stresses the limitations and Incon­ Studwell, W.E.: Subject suggestions 6: Some concerns relaUng to slstencles In slruclure and notation of the UOC. Gives a detailed aerounl of quanllly 01 sUbjects. (Orig. en). Cat.& CJasslf.OUart. Vol.l0, No.4, 1990. the revision policy and procedure. (Author) p.99-104

2-198 42.6-267 2-212 448.6-94 VavllOVB, T.P., Buklna, O.A.: Working tw:hedule. for UDC 678 High StudwelJ, W.E.: Subject suggestion. 5: some concerns relating to art. Molecular Substance. Industry: a tutorial. (Orlg. ru). In: (OI'lg. en). Cat.& Classlf.Quart. VoJ.l0, No.3. 1990.p.91-95 Nauch.kataloglz.llt.1 org.katalogov. �skva, SU 1988. p.76-SO 2-213 446.752 2-199 42.6-33;-36 Vlzlne-Goetz. D., Markey,K.: Subject authority control In onlinecatalog Shamkova, N.B.: Working schedules lor class 55Geology. Geological design. (Orlg. en). In: OPACs and Beyond. Dublln, OH: OGLC Inc. 1989. and Geophysical Sciences: Compilation features. (Orlg. ru). In: p.41-SO, 5 refs. Nauch.kataloglz.f!U org.katalogov. �skva, SU 1988. p.70-75, 12 refs. 2-214 448.753 2-200 42.6-522 Frost, C.O.: Title worda as entry vocabulary to LCSH:. Correlation Kashlda, Y.: UOC 615 Pharmacy: recollections of lis revision. (Orlg. Ja). between assigned LCSH terms In bibliographICrecords with Implications for Joho no Kagaku to GIJutsu. Vo1.38, No.1, 1968. p.58-59 subject access In online catalogs.Cat.& ClassltOJart. Vol.l O. 1989. NO.I-2, p.165-179, 8 refs. 2-201 42.6-622 Sen, B.K.: 05 of U.D.C.Schedule.(Orlg. en). IASUe Bull. Vol.32. No.3. 1 987. 2-215 448.755 p.123-132, 11 refs. Carlyle, A.: Malchlng LCSH and user vocabulary In the library catalog. (Orlg. en). Cal.& Classif.Quart. VoLl0, NO.I-2. 1989. p.37-63. 6 refs. 2·202 42.9; 284 Stoyanka. M.: Adopting UDC schedules (1985 Edition) In a general research library. (Orlg.bg). 81bllotekar 36(1 989)No.4, p.13·15 46 On the Colon ClaSSification See also 2-346

2-216 461 43 On the Dewey Decimal Classification Goplnath, M.A.: Colon Claaslflcatlon: Itsthoory and practice. (Orlg. en). Ubrary Herald Vol.26, No.I-4. 1987/66. p.I-38. 5 refs. See also 2-051 . -110 Presents thenew developments In Ranganalhan's theory of library ctasslfi­ catIon and In his Colon Classification Schedules during 1971-81. Highlights 2-203 43.07 the developments In the concepts of basic subject, personality, matter, Loertscher, D.V.: Book review of Satlla, M.P., Comaroml, J.P.: Inlroduc· energy, space and time. Presents extracts from Ed.7 of the Colon Classifica­ tlOR to the practice o' Dewey Decimal Classification. New York: Envoy tion InCOl'porallng In It the new developments In the universe of subjects and Press 1987. 152p. (Orlg. en). Ubr.& Inform. Sci. Annual. VoL5, 1989. p.91 In the theory of library classlncatlon during 1971-1981.

43.1 2-217 46.4 GOdert, W.: Book review 01 Comaroml, J.P., Satlja, M.P.: Dewey Decimal Chatterjee, A., Choudhury. G.G.: CC7: an evaluaUon of Its development Classification: History and current status. New Deihl: Sterling Pub!. In three planes. {Orlg.en). Ann.Ubr.ScI.& Doc. Vo1.36, No.1 -2. 1989.p.1-27, 1969. VI,176p. (Orlg.en). Int.Classlf. VoLI7, NO.1, 1990. p.35-36, 5 refs. 21 refs.

2-205 43.4 Aman, M., Samore,T.: The Dewey Decimal Classillcatlon: Edilion 20 47 On the Library Bibliographical Classification makes Its bow. (Orlg. en). Wilson Ubr.BuiL Vo!.63, No.6, 1989. p.41-44, 1 ref. (BBK)

2-206 43.6-91 2-216 47.22 Turp, SI.: Using Dewey In a monastic library. (Orlg. en). Sadovaya, A.N.: Perfecting common auxiliaries of race and nationality BuII.Assoc.Brlt.Theol.& Phllosoph.Ubrarles Vo1.2, No.4, 1989. p.14-17, 4 In the LBC. (Orlg. ru). In: Nauch.kataJogiz.llti org.katalogov. Moskva. SU refs. 1988. p.61-94, 13 refs.

2-207 43.6-93, 757 2-219 47.274 SchullZ, L.: DeSigning an expert system to assign Dewey classification Fadlceva, E.N.: The adaptation 01 the Soviet Library-Bibliographical numbers to scores. (OI'lg.en). ln: Proc.Nall.OnlineMeeting, New York, NY, Classification In the GDR. (Orlg. ru). Blbllotekoved. blbtlogr. rubezhom May 9-11 , 1989. Medford, NJ:Learned Inform. 1989. p.393-397. 1 ref. No.12O, 1989. p.53-57, 8 refs. 44 On the library of Congress Classification 2-220 47.469 Sapocnlk, E.: The LBC tables lor children's libraries. (Orlg. ru). Blbllote­ kar' (�skva)., No.2, 1989. p.48-49 2-208 44.6-397 MX Bennett, A.A.: Expansions 01 tho Library of Congress Classltlcallon for 2-221 47.989 Mexico, Class (Orlg. as). Investlg. BlbHotecologlca. Vol. 1, NO.2. 1987. F. Helbig, G.: Das Projeld Klasslflkatlon lOr wlasenschaftllche Blbllothe­ p.51-54, 7 refs. ken. (The project Classification for Scientific Libraries). (Orlg. de). Zbl.Blbl.wes. Vot.104. No.5. 1990. p.213-216. 6 refs. Discussion of considerations concerning a methodology for the establish­ ment of a new classification system for scientific libraries orientedtowards the

118 Int. Classif. 17(1990)No.2 - Classification Literature Soviet library BibliographicalClassification (BSK) combinedwith the advan­ 2·231 647 tages of a subject heading system and a system of auxiliaries. Yerke, T.B.: Preparing forestry terminologies for International compa· tlbllRy In computerized Information systems and databanks. (Orlg. en). In: Kempf, A(Ed.), IntUnlon of Forest Res.Org., Subj.Group S 06.03: 48 On Other Universal Systems Inform.syst.f.forestry·refated subjects: Access, search techno and user needs. Proc. 21·24 June 1986. Birmersdorf, CH 1986. p.19S·203, 5 refs. 485 On CS & T for Libraries 65 On CS and T in the Human Area 5 Seealso2-109, -349 Seealso 2·050, -121 487 On Patent Classification Systems 2·232 65114 NLM'. Unified Medical LanguageSystem. (Orlg.en). Inform. Retrieval & See also 2·328 Llbr. Autom. VoL1S, No.9, 1990. p.7·6

2·222 467 2·233 656 Schwelkhardt, F.: Soma thoughts on hybrid systems especially 'or Baldazzl, A.: Sclenze dell'educazlone e classlflcazlonl blbllograflch •. patant documentatlon.(Orlg.en). WorldPatent Inform. Vol.l l, No.3, 1989. (Education science andbibliographic classification). (Orig. It). Frascatl, IT: p.125·133, 23 refs. Centro Europeo dell' Educazlone 1969. 174p., 241 refs. The cI1apters cover: Epistemological and systematic bibliography of educa­ 489 On Universal CS for Other Special Purposes tionalscience · Class L Education In theLibrary of Congress Classification (Children, Youth, Public Offices, State Documents,etc.) • The Bliss Classification and thework of the Classification Research Group In the field ofeducat lon· The system of Llu Guo and the Influenceof Western Seealso 2·220 systems.

2-223 469 2·234 656 Katemann, Ch.: Ole Systematlk 'Or Klnder- und Jugendblbllotheken. Thomas, A.A.: British Counsellor's views of ERIC terms. (Orig. en). Cat.& Entwlcklung und gagenwArtiger Stand. (Classificationsystems for children's Index., No.95, 1969. p.l ,3·4 and youth Jlbrarles. Development and present status). (Orlg. de). KOln: Assessmentofthe degree of correspondence oflanguage and logic found by Fachhochschule fOr Bibllotheks· und Dokumentatlonswesen 1986. Dlpl0· British counsellors In the Thesaurus of ERIC Descriptors. Some 50 counsel· marbell, Orr.Blbliotheken. lars were addressed of which 50% answered. 22 terms were evaluated. There was disagreement concemlngterms, meanings, and relation between 2·224 469 terms. Formulation of conclusions. Tereslnska, A.: Systems 0' arrangement and notation 0' fiction for children andyoung peopleIn Polish libraries: a proposal for a universal 2·235 656 system. (Orlg.pi). Poradnlk Blbllotekarza, NoA, 1986. p.1S'16 Negrlnl, G.: Novarl, E., Onida, M.: Proposte per '8 reaUzzazlon. dl un thesaurus nel satlore dello sport. (Proposal for the establishment of a 2·225 489 thesaurus In thefield of sports). (Orlg. lt). Romma, IT:CNR CONI Olv.Ooc.e Boyce, B.A., Oouglass,J.SI., Rabalais, L.J.: A sort procedure for the S.... Inform. 1989.83p., 8 refs.:;;Note dl Blbllogr.eDoc.ScI.54. perlntendent of Documents Classlflcallon. (Orlg. en). Chapters cover: Introduction · DeterminatIon of the semantic field of the Governm.lnlorm.Quart. Vo1,6, No.2, 1989. p.175·182, 10 refs. thesaurus · idenllflcalion of thesaurus elements (lexical and semantic) • Formal elements of the thesaurus . Thesaurus examples (In general) . Examples of sportthesauri · References.

6 On Special Subjects CS and T 66 On CS and T in the Socio Area 6

62 On CS & T in the MaUer and Energy Area 2 2·236 663 ROnsches, H.: Thasauruspflegalm Verbunddar Blbllotheken dar ober­ 2·226 629 sten BundesbehOrden.Thessaurus maintenance tn the library network of Tlmotln, A., Ghetaru, S.: The rallonal thesaurus of the terms Included In the German governmental offices. (Orlg. de). Mln.Arb.gem.Parlaments· the InternaUonal Electrotechnlcal Comml&alon's publications. (Orlg. u.Beh., No.67, 1989. p.1·22 en). Probl.lnform.sl doc.Vol,23, No.3, 1989. p.115·124 The thesaurus ofthe German govemmental offices consIsts of eight partial thesauri, viz. for concepts denoting time, historical events, form, persons, 63 On CS and T in the Cosmos & Earth Area 3 subject titles,authorities, geographic Items and subjects. The article descrl· bes theparticular thesauri and demonstrates their specialtiesIn tabular form. 2·227 6376 Snlmshchlkova, I.A.: The effectiveness of use of the ASNTI-gaz compu­ 2·237 666 Van Laer, C.J.P.: Accessibility legal Information by means of shared terized STI system sectoral subject heading list. (Orlg. ru). In: 01 ProbLsoversh.otrasl.slstemy nauch.·tekhn.lnlorm. Moskva, SU 1986. p.62· subJect Indexing. (Orlg. nl). Jurldlsche Blbllothecarls. VoL6, NO.2, 1967. 83 p.17·2O, 6 refs.

64 On CS and T in the Bio Area 4 67 On CS and T in the Production Area 7 2-226 6461 671 Lucas, M.: Mise au point du thesaurus d'une banque de donna" 2·236 Nowak, E.: Classification multidimensional Information In economic statlstlque agricola. (Updatingthe thesaurus for the agrIcultural statistical 01 database). (Orlg. Ir). Paris: InsUtut National des TechnIquesde Documenta· system modelling. (Ong. pi). Pr.nauk AE Wroclawlu, NoA43, 1968. p.Q..14 tlon 1989. sap. 2-239 676 2·229 647 Frleband, S.J.: Burnum, J.: The US Trade Representative (USTR) trade thesaurus; development, Innovation and use. (Orlg. en). inform.Serv.& Schrader, S.: 1st das Oxford·System 'Or den Elnsatz In Computer· Use. Vol,a, No.516, 1988. p.23Q..248, S refs. gestOtzten Systemen gaelgnet? (Is the Oxford·system capable of being used In computer+supported systems?) (Orlg. de). Seea lso 2·1 14 Mllt.Ges.Blblwes.u.Dok.Landbau., No.45, 1989. p.5-16, 13 refs. 69 On CS and T in the Humanities Area 9 2-230 647 Boterot, G.: The muHlllngualevolutlon of PASCAL In relation to lorestry. 2·240 692 (Orlg. en). In: Kempf, A.(Ed.), IntUnlon of Forest Res.Organlzations, Beghtol, C.: Access to fiction: 8 problem In clas.slficatlon theory and SG S 06.03: Information Systems for Forestry·related subjects: Access, practlca, Part 2. (Orlg.en). lnI.Classlf. VoL17, No.1 , 1990. p.21-26, 101 refs. search techniques and user needs. Proc. Birmersdorf,CH 1986. p.17S· 184, 17 refs. Int. Classif. 17(l990)No.2 - ClassificationLiterature 119 2-241 693 thatanaphora play an essential role In human processors' production and Erdetyl, F.: Flexible music cla.alflcatlon echome. Suggestions for the understanding of lexts, Its appropriate recognitionand resolution Is essential extension of thecodes of the scheme. (Orlg. hu). KOnyvtarl FlgyelO. Vol.34, to Information retrieval systems that manipulate natural language texts. The NO.l1, 1988. p.50-67, 18 refs. approaches to anaphora undertaken In theoretical linguistics and NlP are surveyed and the results of research on anaphora as It Impacts on Informa­ 2-242 693 tion processes are presented with specificattention to the detailed studies McKnight" M.: Improving acceaa to music: 8 report 01 the MLA Music conducted at Syracuse University. These studies have provided essential TheaauruoProJoct Working Group. (Orlg. en). Notes. VoIA5, NOA, 1988. baseline data on the extent to which anaphora occur, their likelihood of p.714·721, 12 refs. referrlng to concepts Integral to the topic, their effect on a variety of term­ welghllng schemes,and their Impact on retrleval results. Although the most 2-243 693 effective means of processing anaphora may not have yet been determined, MOiler-Benedikt, M.: Popularmuslk In der Muslkallen-Frelhandsyslema­ It Is suggested that Improved retrieval systems will need to represent the lull Uk - Eln Konzept (Popular music In an open access classification scheme meaning of natural language documents,Including anaphoric references as for musle documents- a plan). (Orlg. de). Forum MuslkblbUothek, No.2, 1 988. well as all other discourse linguistic phenomena. (Author) p.95-120, 21 refs. 2-251 732 2-244 694, 758-94 Gay, l.S., Croft, W.B.: Interpreting nominal compounds for Information Petersen, T.: Information on Images: the Art and Archltecturo Thesau­ retrieval. (O'lg. en). lnform.Process.& Management. Vol.26, NO.1, 1990. rue ae 8 front end for retrieval. (Orlg. en). Proc. Nail. Online Meeting, New p.21-38,22 refs. York, NY, May 9-11, 1989. Medford, NJ:learned Inform. 1989. p.335-341 , II Is sho'>N/1 that simple, knowledge-Intensive algorithms can successfully 6 refs. Interpret many of thecompounds found in technicaldocuments. Experiments Indicate that effective Identification of concepts represf3nted by nominal compounds maybe possible without these knowledge·lntenslve algorithms. 7 Classification and Language (Authors, abbr.) 2·252 733 71 General Problems of Natural language Hahn, U.: TOpic parsing: accounting for text macro structures In full­ text analysis. (Orlg. en). Inform.Process.& Management. Vol.26, NO.1, 1990. p.135-170, 67 Jefs. 2-245 7158-51/4,-68,-62 The rapid proliferation of full-text databases poses serious problems to the Kumagai, N.: Application of Artificial Int8l11genc8. (Orlg. ja). Online natural language processing components Information retrieval systems. Kensaku. VoUO, No.3, 1989. p.l07·116 01 Not taking text-level phenomena of written natura! language discourse Into account causes a marked deaease of perlormancefor many text Information system applications. Consequently, appropriate text parsing faci lities must 72 Semantics be capable 01 recognizing the rich Internal structure of lull· texts on lower levels of text connectivityas well as on the global organlzatloonal level of text Seealso 2-123, -143 coherence. This paper Introduces such a parser which Is based on the conceptual knowledge of Its domain and Is organized as a collection of 2-246 722 distributed lexlcallzed grammar modules(word experts) which communicate lekant, P.A.(Ed.): Word semantics and phrase semantics. (Orlg. ru). through message·passlng. Emphasis Is put on text grammatical specifica­ Moskva, SU: Mosk.obl.pedJn 1989. 123p.Mezhvuz.sb.nauch.tr. tions which state formal. conditions lor recognizing higher-order lext oonstl­ tuenls and their coherent configuration on the global level of textual macro 2-247 723, 396 organization. (Author) Bauer, G., Schneider, Ch.: PAOOK-II: Untersuchungen zur Volltexlpro­ bJernallk und zur Interpretatlven Analyso der Retrlevalprotokolle. 2-253 733 (PADOK-II: problems of full text retrieval and Interpretative analysis 01 retrle· Metzler, D.P., Haas, St.W., Cosle, C.l.: ConJunction, ellipsis, and other val protocols). (Orlg. de). Nachr.Dok. VoIA1, NO.1, 1990. p.21-26, 14 refs. discontinuous constituents In the constituent obJect parser. (Orlg. en). Description of the results of additional tests concerning the question of how Inlorm.Process.& Management. Vol.26, No.1, 1990. p.53·71, 21 refs. much of a document (fulltext versus title + abstract) Is needed for reaching The Consllluent Object Parser (COP) Is a domain Independent syntactic acceptable recall. Results show oonslderable lower recall values for restrlc­ parser developed for use In Information retrieval and similar applications. lis ted document size. The Integration Into the overall methodical concept of purposeIs toextractaslmple hierarchicaldescription ofa phrase or sentence PADOK and first results are the essentials of the Interpretative retrieval that can be used In very general pattern matching procedures to determine protocol anatysls. (Authors) the struclut'al slmllarlty of sentences or phrases that contain equivalent terms. This paper discusses the mechanisms by which COP handles ttle 2-248 723 problems of conjunction, ellipsis, and discontinuous constituents. These lavrenova, O.A.: A knowledge system model used In computerised structures are usually parllcularty difficult to handle In a parser thatdoes not Informatlon.yltam •. (Orlg. ru). Nauchno-tekhnJnform.Ser.2, No.6, 1989. employ domain knowledge or even general semanllc knowledge. COP's p.5-9, 11 refs. mechanisms for these structures are directly tailored for, and, in part, even Descrlbes theexperience and methods of pracllal applicationof a language made possible by, the nature of the Inlended uses of the outputs by the for modelling thesemantics of sclenllflc texts especiallydeveloped to handle Information retrieval matching procedures. (Authors) basic problems In thedesign of llngulslle support for Informationsystems and databases. The formalisation of semantics consists of oomputatlon and 2-254 733 format representation of some actIvity components based on a hypothetical Salton, G., Buckley, Ch., Smith, Marla: On the application of syntactic me­ model of knowledge. (Author) thodologies In automatic text analysis. (Orlg. en). Inform.Process.& Management. Vol.26, No.1, 1990. p.73-92, 32 refs. 2·249 725 This study summarizes various IIngulstslc approaches proposed lor docu· Tatsuml, F., Tetsuya, H., Hlroyukl, K., Ken'lchl, H., Aklo, K.: Architectural ment analysis In Information retrieval environments. Included are standard evaluation of ...mantic network machine. (Orlg. ja). In: Database syntactic methods to generate complex content identlllers, and the use of Mach.and Knowl.Base Mach.: Papers 5th Int.Workshop Database Mach., semantic knowhow obtained from machine-readable dictionaries, and from Tokyo, 1987. Boston, MA 1988. p.544·556, 6 refs. speclally constructed knowlege bases. Certain syntactic term phrase gene­ ration systems are examined In detail and their usefulness for text analysIs purposes Is evaluated. (Authors) 73 Automatic language Processing 2-255 733 See also 2-063, -144 Sembok, T.M.T., Rljsbergen, C.J.van: SILOL: 8 simple logical-linguistic document retrieval system. (Orig. en). fnform.Process.& Management. Vol.26, No.1, 1990. p.1 11-134, 41 refs. 2-250 732 a liddy, E.D.: Anaphor. In natural language processing and Information This paper Introduces loglcaHlngulstJc model of document retrieval sy­ retrieval. (Orlg. en). Inform.Process.& Management. Vol.26, NO.1 , 1990. stems and describes an Implementation of a system callel SllOl.. which Is p.39-52, 27 refs . based on this model. SllOl uses a shallow semantic translation of natural Anaphora Is the discourse-level linguistic phenomenon of abbrevIated sub­ language texts Into a first order predicate representation In performing a sequent reference, pronouns being the most commonly used anaphors. In document Indexing and retrieval process. Some preliminary experiments

120 Int. Classif. 17(1990)No.2 � Classification Literature have been carried out to test the retrieval effectiveness 01 this system. The expectations by Inoorporallng Library of Congress Subject Headings In results obtained show Improvements In the level 01 retrieval effecUveness, online catalogs. which demonstrate that the approach 01 using a semantic theoryof natural language and logic In document retrieval systems Is a valid one. (Authors) 2-265 753 Massicotte, M.: Improved browsable displays for online subject access. (Otlg. en). Inf.Technol.& Llbr .. Vol.7, No.4, 1988. p.373-380, 8 refs. 74 Grammar Problems 2-266 753 Olsen, R.J., Christensen, J.�., Larsen, K.A., Willey, K.: Implementing Seealso 2·113 Nons Keyword/Boolean Searching: a can study. (Orig. en). Call.& Classlf.QuBrt. Vol.l0, No.1-2, 1989. p.125-140, 12 refs. 2·256 743 Endrzej·Boryczka,U.: Automallc recognition 01 grammatical class88 of 2·267 753 worde by the end fragmente. (Orig. ru). Nauchno-tekhn. Inform. Ser.2., Ercegovac, Z.: Augmented assistance In online catalog subject sear­ No.6, 1989. p.24·27, 7 refs. chIng. (Otlg. en). Ref.Llbr., No.23, 1989. p.21-40, 92 refs.

2-268 754 75 Online Problems and Technologies Alt·Kael,H.: Nasr, R., Seo,J.: Immplemonllng a knowledge.based library Informallonsystam with typed Horn Logic. (Grlg. en). Inform.Process.& Seealso 2·059, -064,·069, ·072, ·073, ·124, ·146, ·150, ·151, ·161 , -165, - Management. Vol.26, No.2, 1990. p.249·268, 11 refs. 213, -214, -215, -244 The artIdedlsrussesthe Implementation ofa knowledge base fora library In­ formation system. The system was conceived using a typed logic program­ 2-257 75{ ming language - LOGIN- where type Inheritance Is built In. The knowledge GOdert, W.: Book review of Kemp, D.A.: Computer-baud knowledge base Is structured In a hierarchical taxonomyof library object dasses where retrlevaf. Londoh: Aellb 1988. X,399p. (Orlg. en). lnI.Classlf. Vol.17, No.1, each class Is represented In a frame style knowledge structure and Inherits 1990. p.36-37 the properties of Its parents, and where lnfrastructural lnference rules have been established through typed Horn clauses. (Authors, abbr.) 2-258 751 Matz, P.: Subject aaarching In libraries: present and future, Part 2. (Orig. 2-269 754 en). J.Acad.Llbrnshlp. Vol.14. No.6, 1989. p.l-11 Vyasamoorthy, P.: BasIc program lor sequential search using Boolean operators. (Orlg. en). lnt.lnlorm.Commun.& Educ. Vol.7, NO.2, 1988. p.191- 2·259 751,812 198, 1 ref. Slack, F.: Subject searching on OPACs: a general survey ol laclllU" available on OPACe In academic libraries In the UK. (Orig. en). Vine., 2-270 755 NO.72, 1988. p.8-15, 8 refs. Blazek, A.: Bllal, D.: Problems with OPAC: a case study 01 an academic rauarch library. (Grlg. en). RQ. Vo1.28, NO.2, 1988. p.169·178, 10 refs. 2-260 752 Broadbent, E.: The online catalog: dictionary, classllled, or both? Cat.& 2-271 755 Classl'.Quart. Vol. 10, NO.1-2, p.l05-l24. 26 rels. Klugman, S.: Failures In subject retrieval. (Orig. en). Cal.& Classlf.Quart. The main purpose of the study was to determine If the online catalog can Vo1,10, No.1·2, 1989. p.9-35, 19 refs. function both as a dictionary and classified catalog without requiring addltlo­ Since users seldom thinkof their topiC In terms of library subjecl headings, nal timeor Intellectual efforton thepart of the cataloger. A total of 1842 MARC they encounter problems In finding matches. Online catalogs have helped bibliographic records Hsted In the 370-379 classified section of American somewhatby allowing keyword searching and Boolean combInations.Libra· BookPublishing Record were studied.These recordsdisplayed 2735subJect rlans need to look carefully at the MARC format and at enhanced and headings. or these, 1491 (55%) had a Library of Congress elasslficatlon enriched subject approaches. Machine-generated links and leads can provl· number linked to them. An alphabetical and classified Index was aeated de more paths and wider access to subject Information. (Author, abbr.) using primary SUbjects and theirrelated classification numbers. While such an Index couldbe a useful browsing device IfIntegrated Intoan online catalog, 2-272 755 creating a bona fide dasslfled catalog would require assigning classification Murdock, P.R.: Cataloging catalysIs: toward a new chemistry of cons­ numbers to the secondary subject headings. (Author) cience, communication and conduct In the online catalog. (Grig. en). cat.& Classlf.Ouart. Vol.1 0, No.1-2, 1989. p.65·80, 10 refs . 2-261 752.01 The paper explores the potentialities of an empowered public serviceS! Dale, D.C.: Subject access In online catalogs: an overview bibliogra­ technical services Interface as a liberating and exhilarating vehicle for phy. (OI'lg. en). Cat.& Classlf.Quart.,Vol.l0, No.1-2, p.225-251, 55 refs. change, Improvement, understanding, and access In the online subject The first lIem In this reviewof 55 references Is dated 1978 and thelatest three catalog. (Author) are from 1988. Almost 40% of the Items were published In 1984 and 1985. Slightly over 5O%oftheItems were published In six journals, with the greatest 2·273 755 number appearing In Inform.Technol.& Libraries and lIbr.Resources & Rolland·Thomas, P., Mercure, G.: Subject access In a bilingual online Techn.Servloes. The review Is In three sections: Background section with catalog. (OI'lg.en).Cat.& Classlf.Quart. Vol.l0, No.1-2, 1 989. p.141-l63, 17 general discussionof thetopic and literature surveys; section with material rels. on research anddevelopment; and a lasl secHon dealing with fulure develop­ canadian library records, bilingual by statute, are created according to a ments. common setof rules and standards. SubJect access to the catalogue remains language dependent. Even searches by classlflcallon In some OPACs rely 2-262 752 FR first on an alphabetical Index as an entering key. Fully bilingual OPACs with Jougue!et, S.: Subject access In onllno catalogs: development. In authority control and reciprocal references In one file have yet to appear. France. (Orlg.en). cat.& Classlf.Ouart. Vo1.10, No.1-2, 1989. p.21 3-224 Thus far the Canadian Workplace Automation Research Center has produ· ced ASIRISIRI with an online bilingual thesaurus. The National Library of 2-263 752 Canada has developed within DOBIS a bilingual syslem providing automatic McCarthy, C.: A refere nee llbrarlan'. view al tha online su�ect catalog. linkages between records and authority files. But true bilingual searching (a (OI'lg. en). Cat.& Classl'.Ouart. Vol .l0, No.1-2, 1989. p.203·211, 9 refs. search In one language retrieving records In both languages) could be As truly effective subject retrieval for online catalog users will not bepossible achieved by a built-In automatic translation module.Total accommodation of until the Library of Congress gives a higher priority to consistency In the diacritical marks by terminal keyboards and display monitors Is also a assignment of subject headings - recognizing the Intellectual oomplexlty 01 challenge to be mel. (Authors) LCSH - It Is proposed thatsubject specialistsshould beappointed to monllor the assignment of subject headings al the library of Congress. (Author, 2·274 756 abbr.) Boyce,B.R.,MecLaln, J.P.: Entry point depth and online aaarch using a controlled vocabulary. (Grlg. en). J.Amer.Soc.lnform.Sd. Vol.40, No.4, 2.264 753 1989. p.273-276, 14 refs. Markey, K.: Alphabetical saarchlng In an online catalog. (Orlg. en). The depth 01 Indexing, the number of terms assigned on average to each J.Acad.Llbmshlp. Vo1,14, No.6, 1989. p.353·360, 11 refs. document In a retrieval system as entry points, has a slgnlrlcanleffect on the Analysis of retrieval test data suggest that searchers expect alphabetic standard retrieval performance measures In modern commercial retrieval searches to provide lists 01 assigned subject headings conceptually related systems. Tests 01 the effect of basic Index searCh, as opposed 10 controlled to their topics of Interest. System designers can not respond to their

Int. Classif. 17(1990)No.2 - Classification Literature 121 vocabulary search, In thesereal systems are quite different than traditional Forest Res.Org., SubJ.Group S 06.03: Information systems for forestry­ comparisons of free text searching with controlled vocabulary searching. In related subjects: access, search techniques and user needs. Proceedings. modemcommercial systems the controlled vocabulary serves as a precision Blrmersdorf, CH 1988. p.185-193 device, since the structure of the default for unqualified search terms In these In order to obtaln onHne searchers of high quality the knowledge and systems requires that It do so. (Author) extensive use of terminology Is essential. But even databases covering the same topicsoften differ strongly In thespeclflcterminology they use. Thus the 2·275 756 Increasing number of databases avallable tends to make online searching Van Beveren, Ch.J.: The MEMIC subject headings thesaurus online. more and more difficult. Using some ofDIMDI's databases covering forestrry (Orlg. nl). Open.VO!.20, No.3, 1988. p.92·93, 5 refs. some search aids of theretrleval system G RIPS are presented and explal ned In detall. (Author, abbr.) 2-276 756 Williamson, N.: The role of clas81flcallon In online systems. (Orlg. en). 2-265 758-66 Cat.& Classlf.Quart .. VoUO, No.1-2, 1989. p.95-104, 20 refs. Stone, A.T.: Subject searching In law library OPACs. (Orig. en). Cat.& Classlf.Quart. Vol.l0, No.1-2, 1989. p.181-198, 5 refs. 757 Expert Systems Law libraries that Introdce online catalogs to their publics often experiencethe same kinds of problems and benefits as other types of IIbrarles. Nevertheless certain Issues for OPAC subject searching can beparticularly challenging for Seealso 21-039, -043,-044,-091 law libraries. This paper addresses those Issues, especially the Impact on subject access of the design of 'shared' OPACs and the special problems 2-277 757 Wissen transparent. (Transparent knowledge). (Orlg. de). BIT (FRG). regarding subject searches on government agency names and lilies of laws. Vo1.25, No.4, 1989. p.73-75 DIfficulties with law-related subject headings are also discussed. Finally, keyword and call number searching are examined from the law library

2-278 757 perspective. (Author) Agence Natl. du Loglclel,Groupement de Recherche: Systemes expertset outlla pourl'lntelligence artlllclelle 1989. (Expert systems and tools for 2-266 758-76 artlflclal Intelligence). (Orlg. fr). Vandoeuvre (Campus sclentlfique, Boule­ Rob!nson, M.L: A surveyot SIC code searchable databases on DIALOG. vard des Algulllettes, BP 239, 54506): ANL 1989. 242p. (Clrlg. en). Online Review. Vol.13, No.4, 1989. p.313-319, 3 refs. Standard Industrial Classification (SIC) codes are a type of controlled

2-279 757 vocabulary used by business searchers when searching for corporate and Cavanagh, J.M.A.: Library applications of knowledge-based systems. Industry data. Two four-dlgltverslons of the SIC coding scheme are used on (Orlg. en). Ref.Ubr., No.23, 1989. p.1-19, 76 refs. business databases available through DIALOG. Some DIALOO database A broad range of library applications for expert systems Is identified. Descrip­ producers have opted to create a variation of the codes for use In their tions of developments Include systems for document selection, cataloguing, products. With the Impending e!ght-dlglt SIC coding scheme from Dun and to classification, reference, retrieval, and Instruction. (Author, abbr.) Bradstreet, It will become Increasingly difficult for the DIALOG searcher know which version of the codes to use when search ing a particular database. A survey of the use of SIC codes In business databases on 2-280 757 Hamza, M.H.(Ed.): Expert systems: theory and applications: lASTED DIALOG was undertaken. This paper presents the findings of this survey. Int.Conf., GGnev., lASTED Int.Conl., Geneva,June 16-18, 1987. ProCH­ (Author) dings. (Orlg. en). Anahelm,etc.: Acta Press 1987. 11,228 p. ISBN 0-88986- 118-8. 2-267 758-97 Some of the papers of the 52 presented have been Included under different Agger, St., Jensen, H.: Th. BOOK HOUSE · visual design. (Orlg. en). subject groups of this Issue. Problems discussed Include, e.g. conceptual Roskllde, OK: RlsO Natl .Lab. 1989. 45p;, 69 l11ustr. modelling,producllon rules modelling,deductive databases, object-centered The reportIs based on thecontribution to the project from the Royal Danish knowledge representation, AI languages, etc. Art Academy's Architect School. It desCllibesthe approach used to cope with the problems of conveying through a computer-based medium the visual Information necessary to support and enhance user Interaction with the 2-281 757 Travis, I.L: Knowledge-basad systems In Information work: a view of system. the futur•. (Orlg. en). Ref.Llbr., No.23, 1989. p.41-60, 20 refs. 2-268 758-92 Pejtersen, A.M.: The automaticbook machine In HJortesprlng.(Orlg. da). 2-282 757 Vickery , A., Brooks,H.M., Roblnson,B.A., Stephens,J., Vlckery,B.C.: Expert Blbllotek. Vo1.70, NO.l0, 1988. p.347-350 system for ref.rral. (Orlg. en). London: Brlllsh Library 1988. 233p. '" Ubr.& Inform.Res.Rep.No.66. ISBN 0-7123-3146-8. 2-289 758-92 Describes the development and evaluallon of PLEXUS, an expertsystem Goodstein, LP., Pejtersen, A.M.: The BOOK HOUSE: system functiona­ lity and evaluation. used to aid microcomputer retrieval of referral sources. The function of the (Orlg. en). Roskilde, OK: AlsO Natl.Lab. 1989. 215p. system Is to give Ineperlenced users the kind of aid that a skilled search ISBN 87-550-1537-9. Intermediary would provide. The system helps to formulate a search, create The BOOK HOUSE (and Its predecessor, the BOOK AUTOMAT) are a search strategy, modify the search strategy as required, and present the computer systems aimed at supporting library users In finding fiction. They results of the search. To carry out these taks the system uses the techniques are examples of systems using a cognitive work analysis as the basis for of artificial intelligence - knowledge representation, production rules, Inferen­ design and evaluatlon- and build on earlier studlesof the Information retrieval ceand control, a natural language Interface, and some simple user modelling task using actual user-librarian negotiations. These studies lead to the devices. (Authors, amended) identification of a set of user strategies for retrlevlng literature as well as a multi-dimensional framework for classifying fiction. Both of these have been 2-283 758-26 necessary prerequisites for thedesign of the BOOK HOUSE,a flexible online Ledwith, RH.: Development of a large, concept-oriented database for data base retrieval system for novice users - both children and adults. The Information retrieval. (Orig. en). In: 11th Int.Conf.Res.& use of icons within a spatial metaphor to facilitate directmanipulation of the Dev.lnform.Relrleval, Grenoble, 13-15 June 1968. Grenoble, FR: 1988. displayed objects forms the basis for the Interface In supporting the users' p,651-661 ,15refs. navigation in the BOOK HOUSE. The design and evaluation of the prototy­ The development of concept-oriented databases using AI knowledge repre­ pe systems are covered In this report. sentation schemes is proposed as a step towards Improving the precision and recall of IR systems. Currenlly underway is the augmentation of a 2-290 758-92 238,000 citation database, Chemical Abstracts (CA) Volume 105, by addition Pejtersen, A.M.: The BOOK HOUSE: modelling users' needs and search of detailed conceptual Information In the form of frames and hierarchies. The strategies as a basis for system design. (Orlg. en). Roskilde, OK: RlsO Nall.Lab. 1989. 112p., 76 refs. Inilial text data Is parsed using natural language processing (NLP)techniques 10 create frames describing the semantics of Ihe Index enlrles In the See abstract above (2-289). The cognitivework analysis which preceeded database, with theslots In the frame being pointers Intoa very large semantic the design and evaluation of the prototype systems are covered In this report. network of conceptual objects (956,000 objects) . To examine the resultant knowledge base, a simple hypertext system Is proposed, with theconceptual 2-291 758-992 Information serving as pathways to connect related citations. (Author) Carr, A.R: Strachan, N.: Development of. system for treatment 01 Blblo headingsIn an OPAC system. (Orlg. en). Cat.& Index., No.95, 1989. p.5-

2-284 758-47 6 Kaiser, D.: Terminological search Instruments online. Some examples using the retrleval language GRIPS. (Orlg. en). In: Kempf, A.(Ed.), Int.Unlon

122 [nl. Classif. 17(1990)No.2 - Classification Literature 76 Lexicon/DictionaryProblems compression and standardization In sc!entlflc publications). (Orlg. en). fnl.Classlf. Vol.17, No.1, 1990. pA2-43 Seealso 2-049, �1 11 78 Subject-Oriented Tenninology Work 2-300 762-' 766' 793 Belyaeva, LN.: Designing linguistic Information databases fo r co�pu­ 2-311 78·628 terlsed text processing systems. (Orlg. ru). Stat.rechl I Sprachendlenstd.BM f.Post u.Telekommunlkatlon: Termlnologle der Ver­ , aVlomat.pererab.tekslov, No.7, 1988. p.8-28 tragawerke der Intemallonalen Femmeldeunlon. Oan', de - fr - en. (TermInology olthecontracts of theInternational TelecommunicationUnion) 2-301 763 (Orlg. de, fr, en). Termlnol.et Traducllon, No.3, 1989. p.231-288 Pollkarpov, A.A.: Bushuev8, C.V.: An experiment In constructing a con­ Glossaryof the384 most Importantterms (telecommunlcallon,conferences, textual dictionary and analysis of Its structure. (Orig. ru). In: Teor.1 contracts, administration, etc.) are conlaltled in the Internallonal Telecom· prakt.probt.prlkl.lingv. Moskva, SU 1988. pA0-65,48 refs. municatlon Treaty (Nairobi 1982) of the International Telecommunication Union (a specialorganIsation of theUNO)and In lIS bylaws and recommen­ dations. 77 General Problems of Terminology 2-312 78-63 Seealso 2·045, ·348 Moos, A.: Tormlnologlearbelt In dar Schelzerischen Bundesverwal­ tung. (Terminology work In the Swiss Federal Administration). (Orlg. de). Termlnotet Traduction., No.3, 1989. p.149-154 2·302 77.961 Schmitt, P.A.: Tarmlnologlscho Dlplomarbelten. RlchtllnJen des Fachbe­ 2-313 78·82 falchs Angewandte Sprachwlssenschaft In Germershelm. (Terminological Atanas!u, P.: Teodoru,V.: Aspectsof terminology In the field of Informa­ theses. Guidelines of the Department for Applied Linguistics In Germershelm tion and documentation. (Orlg. ro). Blblloteca, NO.4, 1989. p.17-18t 9 refs. - University 01 Malnz). (Drlg. de). Lebende Sprachen. Vol.34, NO.4, 1989. p.153-159, 15 refs. 2-314 78·82 Guidelines concerning theme selection, acquisition of material, outlines, BrOOmeler, B.: Ole Phllosophle der TID sowle die Art Ihrer Darstellung. formal design and Information offer. They are based on experiences and (Conceptionof the TID and knowledge gained from more than300 terminological theses, submitted in the Its way of representation). (Ong.de). In: Deutsche Gesellschaft fOr Dokumentatlon: Jahre DGD - Perspekllve Information. 40 years of existence of ADIIF AS (Fachberelch Allgemeine Sprachwlssen­ 40 Dokumenlartag Aachen 1988. Frankfurt 1989. p.153-162 schaft).

2-315 78-8216 2-303 772 Hoeflg, W.: (TIle database KOMINTERM). Sager, J.C.: Term formation. (Orlg. en). Lebende Sprachen. VoL34, NoA, Ole Datenbank KOMINTERM. (Orlg. de). In: Deutsche Gesellschaft Dokumentation: 40 Jahre DGD - 1989. p.159-161 fOr Perspektlve Inlormallon. Dokumentartag Aachen,27.-2 9.9.1988. Frankfurt Distinction between primary and secondary term formation. The former ac­ companies concept formation and Is �monolingual", proVisional, accompa­ 1989. p.174-181 Information on theterminology databank for thearea of communication and nied by a stlpulatlve or otherwise temporary definition. The latter occurs when Inlormatlonwhlchhas been established with the software system ALLEGRO a new term Is aeated for a known concept and happens as a result of In the State Library of Prusslan Culture and which Is being maintained there monolingual revision of a terminology. The article extends on the characte­ ristics of the two kinds of term formation. and used for retrieval purposes. The database consists of some 20,000 dataselsand Indudes under 'information' theconceptual areas of archives, library sdence, documentation, Information and other concepts. Under 2-304 772 'communIcation' the concept areas are covered of publishing, printing, Soroklna, N.G.: Modalllng word formation semantics In computerised Journalism, broadcasting and television, movies and audiovisual media as Information system vocabularies. (Orlg. ru). Sb. nauch. trlLenlngr. Inst. well as telecommunlcallon. The sources for the datasels are desalbed and kult., No.116, 1988. p.55-S9, 10 refs. the role of the Commlnee on Terminology and Linguistic Problems (KTS) (of the German Documentation Society) In maintaining and developing the 2-305 773 databank Is explained. Neuhaus, J.-P.: Le termlnologue au Hln d'une enlreprlse du secleur prlv' ". Les exlgences d'engagement, ses domalnes d'acUvlt6. (The lerml­ 2-316 78·822 nologlst In acompanyolthe private sector.(Orlg. Ir). Termlnol.etTraduc!lon., Ermert,A.: Ookum.ntatlon und Dokumenttelle und Ihre Rolle lnderTID. No.3, 1989. p.161-175 (Types and partsof documentsand their role In TID). (Orlg. de). In: Deutsche Gesellschaft fOr Dokumentatlon: 40 Jahre DGD - Perspektlve Information. 2-306 776 Dokumentartag Aachen, 27.-29.9.1988. Frankfurt 1989. p.186-193 Irazazabal, A.de, Alvarez, S., Zarco, J.: EURODICAUTOM - the Commis­ The "Standard Generalized Markup Language� (SGML) (ISO 8879) which sion 0' the European Communities term databank. (Orlg. es). serves the Input of texts Is explained and thesignificance of the recognition Rev.esp.doc.dent. Vo1.12, No.1 , 1989. p.1'-12, 4 refs. of kinds and parts of documents Is stressed especially regarding 1) the publication process, 2) the bibliographic deSCription and 3) the subject 2-307 776 analysis and evaluation. Also explained are the different terminological Sotlropoulos, M.Th.: Conceptual-malhemallcal modela as a brldge lrom structures the fields of kinds and parts of documents In the International the general theory ofterminology 10 Informatica. (Orlg. en). Termlnol.et 01 standard ISO 5127/2 and In theGerman standardlzallon project DIN 31 631. Traduction., No.3, 1989. p.135-148, 8 refs. The article is composed of the lollowlng chapters: DeflnWon and representa· 2-317 78-8227 tlon of knowledge - Knowledge processing - Semantic networks - Compute· Lalslepen, K.: Audlovlsuelle Oakumenl•. (Audiovisual documents). (Ong. rlzed termlnography - Kinds and functions of terminological data banks - de). In: DeutscheGesellschaft fOr Dokumentatlon:40 Jahre DGD - Perspek­ Knowledge data bases - ExpertSystems. live Information. Dokumentartag Aachen, 27.-29.9.1988. Franklurt 1989. p.163·173 2-308 776 Machova, S.: Mlnlhofer, 0.: Developmenl of a mullliinguai term data­ 2-318 78-834 bank. (Orlg. cs). Vyber.lnf.organ.avypoc.techn. Vo1.22, No.2, 1989. p.225- Barkow, G., Hesse, W., Klttlaus, H.B., Luft, A. et al.: BegrUfilche Grundlaw 235 gen fOr die 'rOh.n Phasen der SoHwar.Entwlcklung. (Conceptual bases for the early limes of software development). (Orlg. de). Information 2-309 7768-73 Management. Vol.4, No.4, 1989. p.54-60, 15 refs. Aleksandrov, LV., Bukhalkln, Yu.M., VoznesensklJ, V.V.: Establishing a knowledge bank In an engln.. ,lng field. (Orlg. ru). Nauchno­ 2-319 78-962 tekhn.1nform.,Ser.1., No.6, 1989. p.16-17, 12 refs. Fosken, O.J.: Book review 01 Riggs, F.W. (Ed.): Ethnlclty. Inlereacts Glossary. Concepts and terma used In ethnlclly research. (Av8.ilable 2-310 778 from:) Indeks Verlag, Frankfurt, XXX+2OSp. (Orlg. en). Int.elass!f. VoI .H, Bles, W.: Book review ol lnformatlonsverdlchtung und Standardlsl8- No.1, 1990. p.43·44 rung In wla8Gnschaftlichan Ver6ffenlilchungen. Wlss.Z.Ka/1·Marx­ Unlv., leipzig, Ges.wlss.Relhe 37(1988)No.6, p.521-623. (Informallon

Int. Classif. 17(1990)No.2 - Classification Literature 123 79 Problems of Multilingual Systems and 2-332 845 Romanluk, E.: Macintosh-baaed keyword aceessto IEEE serial publica­ Translation tions. (Orlg. en). Inlorm.Technol.& llibr. VoL7, No.4, 1988. pA11·416

See also 2-300 2-333 847; 812 Larina, V.G., Khorosheva, T.V.: catalogue maintenance In USSR state 2-320 791 archlvea. (

81 General Problems of Applied C & I Seealso 2-089

Seealso 2-333, 2-259 2-335 857 Korydnskl, C .• Newell.A.F.: Natural-language processing and automatic 2-321 812 Indexing. (Orlg. en). Indexer. VoL17, No.1, 1990. p.21-28, 57 refs. Chlzhkova, G.I., Chortjunova,L.V.: On systematic catalogues. (Orlg. (u). The task of producing satisfactory Indexes by automatic means has been Sov.blbllotekoved., No.3, 1989. p.53-56, 16 refs, tackled on two fronts: by statistical analysis of text and by attempting content analysis of the lextIn much the same way as a human Indexer does. Though 2-322 814 statisHcaI techniques have a lot to offer for free-Iexl database systems, Junglnger. F.: Regeln fOr den Schlagwortkalalog: RSWK. ErgAnzungan neither method has had much success with back-of-the-bookIndexi ng. This und SerichUgungen. (Rules for the Subjecl Catalogue: RSWK. Extensions review exami nes some problems associated with the application 01 natural­ and corrections). (Orlg.de). 8ibllotheksdlenst. Vol.22, No.6, 1988. p.552-563 language processing techniques to book texIs. (Authors)

2-323 814 Scharmann, U.(Comp.): Deutsches BibliothekslnstilUl: Anwendung der 86 Indexing of Secondary Literature Regeln fOr den Sehlegwortkatalog (RSWK) In 611entllchen Blblloth.. ken, (Appllcatlons of theRules for theSubject Catalogue (RSWK) In public 2-336 libraries). (

2·325 818 2·338 Kalczynska, M.: Tho form 01 tho subject catalogue In Lwow University 864 Todorov, R.: Representing 8 scientific field: a blbllometrlc approach. LIbrary Intholnter-waryeara.(

2-326 818 Khakhaleva, N.I.:Terma revealing thestructure of a large l/brary collec­ 87 C & I of Non-Book Materials lion. (

B4 C & I of Primary Literature (except 85) 2-340 878 Smllher, R.: 'Is Britannia a personality?': 80me questions arising while Seealso 2-193 Indexing the Imperial War Museum's collection. (Orig. en). Indexer. VoL 17, No.1,1990. p.7-11 2-328 842; 487 Illustrates - even li lt does not resolve - some of the problems In providing SlulSkln, A.A., Marshak, N.A., Barlsova,I.N., Lukanln, VA: Increasing the subject access to the extensive, International, multi-media collections of the efleetlvene" o. patent search baaed on bibliographic description. Imperial War Museum. Chief among Ihese Is the variel)' of ways In which (Orlg. ru). Nauchna-.tekhn.lnform., Ser.1, NoA, 1989. p.11-13, 4 refs. Items In museum collections can be relevant to a particular line 01 enquiry. (Author) 2-329 844 Gee, A.: Newspaper cuttlnga: alternative to Indexing? (Orlg. en). Local Studies Librarian VoL7, NO.2, 1988. p.7-11 BB C & I in Subject Fields 2-330 844, 977 Nelson, D.W,: Newspaper Indexing with microcomputers: the Washing­ 2-341 88-39; 758-39 ton Statoexperlence. (Orig. en). Reference ServicesReview. Vol.17, No.4, Hili, l.l.: Geographic Indexing lor bibliographic databases. (Orlg. en). 1989. p.91-96 Resource Sharing & Inl.Network. Vol.4, No.2, 1989. p.1-12, 14 refs.

2-331 844 NI 2·342 88-51/4 Okorafor, E.E.: Newspaper Indexing In Nigerian libraries. (Orlg. en). Berrul, C.: Indexing medical reports: The RIME approach. (Orlg. en). Indexer. VoU7, No.1, 1990. p.35-38, 14 refs. Inlorm,Processlng & Management. Vol.26, NO.1 , 1990. p.93-109, 16 refs, Few Nigerian libraries Index dally and weekly newspapers. Proposals for The paper focuses on Ihe RIME system aimed at the Indexing of medical acHons towards establishment of standards and authorlty"lIsts. reports In a multimedia environment. This particularappllcalton Is viewed as

124 Int. Classif. 17(1990)No.2 · Classification Literalure representative of a large set of stiliunanswered needs of large communities 93 Organization of C & I on a National and of users: Domain expertsdealing withonline spedallzeddocumentation such as software engineers, medical specialists and soon. In thepresent applica­ International Level tion case, textual Information appears as an Interesting medium to access related pictures In the database. After thepresentation of the application and 2-350 934 SU a study of theparticular corpusInvolved a semantic model Is defined for the VanSkaja, G.P.: Gavrlllna, N.V., !vanova, O.A.: Onee more on centralized documentswhich Is based on a Conceptual Language. Thelndexlng process cataloguing. (Orlg. ru). Sov.blbllotekoved., No.3, 1989. p.49-53, 3 refs. Is detailed as well as Its vartous linguistic components which perform the Proposal to let librarians do the centralized dasslficatlon of books as the translation of every medical report accord ing to this semantic model. publishing houses seem to fall to do this properly. At the same time catalog (Author, amended) cards could begenerated with the notations of the BBK and the UDC.

2-343 88-92 BrOnlng, Ch.: SchiagworterachlieBung von BeUetrlstlk In OHentilchan 95 Education and Training in C & I Blbllothekan. Elne Untersuchung der Methoden unter besonderer BerOck­ slchllgung der RSWK-Anwendung. (Indexing of belles-lettres In pUblic libra­ Seealso 2-302 ries. An Investigation of methods with special regard of an application of the (Orlg.de). KOln: Fachhochschule Blbllotheks- und Dokumentatlons­ RSWK). 2-351 951 wesen 1988. Dlplomarbelt OOenti.Blbllotheken. Fugmann, A.: Unkenntnla • unsar Schlckaal. (Lack. of knowledge - our fate). (Orlg. de). Mln.bI.Ges.Dt.Chemlker, No.16, 1989. p.70-73 2-344 86-94 Pleads for therecognition of thenecessity of subject knowledgefor a correct Roberts. H.: 'Do you have any picture. of... ?' sUbJect accesa to works functioning of procedures. of art In visual collectfons and book reproductions. (Orlg. en). Art DoctJmentation.Vo!.7, No.3, 1988. p.87-00, 6 refs. 2-352 952 Kovacs, B.: An educational challenge:Teaching cataloging and classi­ fication. (Orlg. en). Ubr.Resources & Techn.Serv. Vol.33, No.4, 1989. p.374-381, 33 refs. 9 Classification Environment In recent years artldes have appeared In the literature on the changing role of thecataloger, and complaintshave arisen about the comprehensiveness of cataloging and dasslficallon courses offered In graduate library schools. 92 Persons and Institutions This paper discusses the challenges of offering such courses and In presenting all the necessary Information In Introductory cataloging and 2-345 924 classification courses for M.L.S. students. (Author) The Margaret Mann Cnatlon, 1989: Lois Mal Chan. (Orlg. en). Ubr.Resources & Techn.Serv. Vo!.33, NoA, 1989. p.421-423 98 User Studies 2-346 924 Suklasyan, E.A.: S.R.Ranganathan's school and our times. (Orlg. ru). 2-353 986 Blbllotekoved.l blbllog. za rubezhom, NO.122, 1989. p.1 03-107, 115, 20 refs. Craver, K.W.: The Influence of online catalogs on academic library use by COllege-boundhigh schoolsenior •. (Orlg. en). RQ. Vo1.28, No.2, 1988. 2-347 925.313 p.22O-231, 13 refs. Beaujean, M.: Kommlulon d.. Deutschen Blbllothekslnatltut. fOr SacherschlisBung.TAtl gkeltsberichte 1987. (Commission for SUbject Inde­ xing of the German Library Institute). (Orig. de). Blbllotheksdlenst, Vo1.22, No.4-5, 1988. pA12-414 99 Standardization in C & I

2-348 925.77 Seealso 2-053 Port, P.: Ole Arbon.wel.edes KTS und Ihre Auswlrkungen auf die Arbelt 2-354 anderer Gremlen. Nutlung und Zitlerung der TID. (The working methodof 995 KTS (Committeeon Terminologyand Language Problems) and Its Impact on LINCE: a proposal for standardisation of database access languages. the work 01 other workinggroups). (Orlg. de). In: Deutsche Gesellschaft far (Orlg. pt). Clenc.lnform. Vo!.17, No.2, 1988. p.141-145 Dokumentatlon: 40 Jahre DGD - Perspektlve Information. Dokumentartag The Brazilian Institute of ScientlficandTechnologlcal lnformatlon (IBleT) and the Nuclear Information Centre of the National Commission for Nuclear Aachen, 27-29 Sepl.1988. Frankfurt 1989. p.1 82-1 85 Energy are developing a language, LINCE, which will be used for Input to 2-349 926; 485 different computers. II operales through commands and wlll be used mainly BOllmann,E.: Erat ..TreHen der LIBER CONSPECTUS Group In London. for handling referral and bibliographic data. (Orig.abstt.,abrldged) (First meeting of the L1BER CONSPECTUS Group In London). (Orlg. de). 2-355 Mln.Verelnlg.Oslerr.Blblloth. Vol.41 , No.2, 1988. p.17-19 996 The first meeting was held In May 88. Thegroup promotes theUse of the Chan, L.M.: A subject cataloging code? (Orlg. en). Cat.& Classif.Quarl. CONSPECTUSclassification scheme. VoL l0, No.1-2, 1989. p.199-202, 4 refs. Tries toanswer the 4 questions: 1) What Is a SUbject cataloging code? 2) Do we need a subject cataloging code? 3) Is II economically feasible to develop such a code at this point In lime? and 4) Who will develop It?

2-356 997 Gavin, P.: La collaboration dana l'lndexation rnatlillres. (Collaborallon In subject Indexing). (Orlg. fr). Hors-texte, No.23, 1987. p.5-7 Pleads for the establishment of International standards for subjectcataloglng.

Personal Authors

Aderlblghe,M.R. 2-210 Anderson, J.D. 2-196 Barkow, G. 2-318 Bellamy, L.M. 2-121 Agger, SI. 2-287 Andr6, J. 2-169 Bates, M.J. 2-161 Belyaeva, L.N. 2-300 Aguirre, J .L. 2-092 Atanaslu, P. 2-313 Bauer, G. 2-247 8enedlktsson, D. 2-096 Alt-Kacl, H. 2-268 Baumann, E.V. 2-095 Bennett, R.A. 2-206, 2-209 Aleksandrov, L.V. 2-309 Baker, B.etsy 2-077 BeauJean, M. 2-347 Berrut, C. 2-342 Aman, M. 2·205 Baldazzl, A. 2-233 Beck. H. 2-046, 2-001, 2-056 Barsteln, L. 2-067 Anderson, Ch.R. 2-079 Bara, I. 2-112 Beghtol, C. 2-240 Bertrand-Baschwltz 2-032

Int. Class if. 17(1990)No.2 - Classification Literature 125 Bertrand-GastaJdy,S. 2-163 Hamza, M.H.(Ed.) 2-260 Moos, R. 2-312 Sparck Jones, K. 2-065 Bestougeff, H. 2-091 Harter, St.P, 2-082. 2-062 Mailer-Benedikt, M. 2-243 Srinivasan. p, 2-154 Biellcka.l,A, 2-1 17 Helner-FrelUng. M, 2-053 Murdock, P,R, 2-272 Stone. A.T, 2-285 Bles, W, 2-141. 2-003, 2-310 Helbig, G. 2-221 Studwell, W.E. 2-21 1. 2-212 Blru, T. 2-152 Henderson, K.l. 2-022 Nath, P. 2-188 Suklasyan, ER. 2-346 Biswas, S.C, 2-002 Hildreth, Ch.R. 2-075 Neale, I.M, 2-043 Surleau, L 2-170 Black, W,J .2-042 HIII,LL 2-341 Negrlnl. G. 2-235 Svenonlus, E. 2-1 16 Blake, V.l.P. 2-339 Hoeflg, W. 2-315 �elson, D.W. 2-330 Swan HIIl,J. 2-034 Blazek. R. 2-270 HOgemann, S. 2-182 Nelson, M.J, 2-153 SWeeney, R, 2-023 BOllmann, E. 2-349 Horvathne. Z. 2-1 14 Neuhaus, J.-P. 2-305 Szpakowska, J.-K. 2-031 Boterot, G. 2-230 Huang, J.S. 2-130 Ngah, ZaJnab Awang 2-177 Bourret, A. 2-181 Noreault. T, 2-081 Takahlro, A. 2-148 Boyce, B.A. 2-225, 2-274 II'ln. V.V. 2-104 NOWak, E. 2-238 Tatsuml, F. 2-249 Bratcher. P. 2-028 Irazazabal, A.de 2·306 Taylor, H. 2-337 Broadbent, E. 2-260 Iyer. H, 2-113 Ochmanska, A 2-109 Tazl, S, 2-171 8rooer. A,J, 2-129 Okorafor, E,E, 2-331 Tereslnska, A, 2-224 Broomeler, B. 2-314 Jakowlewltsch. D, 2-105 Olsen, R.J, 2-266 Thomas, AR. 2-234 Bruning, Ch. 2-343 James, SI. 2-035 Onodera, N, 2-164 Tlllmanns,St. 2-334 Brzozowski. J,P, 2-071 Johansen, Th, 2-122 Tlmotln,A. 2-226 Buchan, R.l. 2-1 11 Jones, LP, 2-156 Palce, C.D. 2-147 Todorov, R. 2-338 Bunlyalov, AR 2-106 Jouguelet, S. 2-262 Panevovli, M, 2-160 Travis, I.l. 2-281 Burger, E, 2-004 Junglnger, F, 2-322 Pao, M.l. 2-189 Trickey, K, 2-021 Burgess, M. 2-025, 2·026 Pedrycz, W. 2-134 Turp, SI. 2-206 Kaiser, D, 2-284 Pejtersen, A.M, 2-288. 2·290 Carlyle, A. 2-215 Kalczynska. M, 2-325 Perry.SA 2·151 Van Severen. Ch.J. 2-275 Carman, C.S. 2-133 Karbach, W. 2-098 Petersen, T. 2-244 Van Laer, C.J.P. 2-237 Carr, A,A, 2-291 Kashlda, Y. 2-200 Plalsant-Swenn, K. 2-166 Van lIew, H.D, 2-173 Cavanagh, J,M,A, 2-279 Kalemann, Ch, 2-223 polikarpov.A.A, 2-301 Vanskaja, G.P, 2·350 Cawkell, A.E, 2-338 Khakhateva. N,1. 2·326 Porter, M. 2-069 Vavllova, T,p, 2-196 Chan, l.M, 2·355, 2-190 Kimolo, H. 2-157 Port, p, 2-348 Verdooth .A, 2-037 Chane�99, A. 2-217 Kinnell, S,K, 2-165 Vickers, J.A. 2-047 Chlzhkova. G,I. 2-321 Klshlda, K, 2-138 Qulrchmayr. G, 2-040 Vickery, A. 2-262 Chowdhury. G,G. 2-125 Klugman, S. 2-271 Vlzlne-Goetz, D, 2-213 Chu,e.M, 2-186 Kocere, W.Ya, 2-184 Ramalah, l.S. 2-140 Vyasamoorlhy, p, 2-269 Clark, B.M ,2-067 Koryclnskl. C. 2·335 Rapp, R. 2-143 Clark, Sh. 2·085 Kovacs. B, 2·352 Rees-Pone,. 1.K. 2·119 Walke" St. 2-064 Cleverdon, C. 2-060 Kozlova. l.V. 2-123 Rhodes, S. 2-149 Walson, P,D. 2-066 Cochrane, P .A. 2-057 Knilova, A. 2-137 Rlesthuls. G. 2-194 Weldenfeld, G. 2-168 Comlesse, X, 2-103 Krause, J, 2-193 Risko, A. 2-1 15 Willett, p, (Ed.) 2-059 CousSQulet, J.-P. 2-127 Kumagai, N. 2-245 Ritzier, C. 2-160 WilliamsOfl, N. 2"276 Craer, K,W, 2-353 Robertson, S.E, 2-126 Williams, D,E, 2-178 ereth. Sh,D. 2·074 lalslepen, K. 2-317 Roberts, H. 2-344 Wittmann,C. 2-185 Croft, W.B, 2·066 Lamonde, Y. 2-1 10 Robinson, M.L. 2-266 Woodard, B.S. 2·084 Lang, R et aJ.(Eds) 2-007 ROttsches, H. 2-238 Woods, WE 2-008 Dack, D, 2-051 laresse. E. 2·058 Rolland-Thomas, P. 2-273 Woodward. T. 2-108 Dale, D.C. 2-261 larlna. V.G, 2-333 Romanluk. E, 2·332 Derham. l.D,(Comp,) 2·024 Laske, O. 2·099 Rothenberg, D. 2-060 Yannakoudakls, E.J, 2·044 DI Manzol. M, 2·097 lavrenova, nA. 2-246 Roth, D.L. 2-076 Yerke, T,B, 2·231 Dickstein, R ,2·027 lazlnger,8.S, 2-120 Rust,A, 2-049 Diodato, V. 2-078 Leduc, 8, 2-030 Zarrl,G.P, 2-041 Ooszkocs,T.E. 2-072 Ledwith, RH. 2-263 Sadovaya, A.N. 2-218 Dykstra, M, 2-150 legrand, J. 2-100 Sager, J,C. 2-303 lekant, P,A.(Ed.) 2-246 Salomon, K. 2-107 Endrzaj·80ryczka,U. 2-256 lennon, M, 2-063 Salton, G. 2-061 , 2-070, 2·254 Ercegovac, Z, 2·267 Lenz, A. 2·132 Sapocnlk, E. 2-220 Erdelyl, F. 2-241 Liddy, E.D. 2-250 SaH)a, M.P. 2·135 Ermerl, A. 2-316 Llndlner, I. 2-142 Scharmann, U,{Comp,) 2-323 Llu, Songqlao 2·136 Scherrer, J.A. at al 2-050 Fadlceva, EN, 2-219 LI, We] 2-172 Schmldl, J.M, 2-175 Fagan, J.L. 2-158 loerlScher, D,V. 2-203 Schmitt, P.A. 2-302 Fadoseeva, TV. 2-145 Lucas, M. 2-226 Schrader. S, 2-229 Ferre. A-G. 2-006 lu, Xln 2-124 Schultz, L. 2-207 Fldal, R. 2-163 Schulze, Hans H. 2·005 Font, J.-M. 2-093 Macleod, LA. 2-146 Schwandt, J. 2-195 Foskett, D,J. 2-319 MaChOV8 , S, 2-308 Schwarz, Ch, 2·144 Fox, EA 2-162 Malbec, C. 2-324 Schwelkhardl. F. 2-222 Frakes,W,B. 2-066 Mamede, H,C,A, 2- 116 Scon. sl. 2-159 Franklin , J. 2-327 Markey, K. 2-264 Sembok,ToM,T. 2-255 Frleband, S,J. 2-239 Marlolh, H, 2-094 Sen,B.K. 2·201 Frost, e.D. 2-214 Maruyama, Sh. 2·192 Shamkova, N,B, 2-199 Fugmann, R. 2-351. 2-046 Maruyama,Sh, 2-191 Shaw, M.L.G, 2-101 Massicotte, M, 2-265 Shlaw-Dong, D.L. 1-131 Gamat-Eldln, M.S. 2-128 Mater, E. 2·054 ShlmamurB, T, 2-167 GavIn, P. 2-356 McCarthy. C. 2-263 Shlmomura, RH. (Ed) 2-029 Gay, L.S. 2-251 McKnight" M, 2-242 Shulchl. U. 2-176 Gee, R, 2-329 Meadow. Ch.T, 2-063 Slack, F. 2-259 Giger. H.P. 2-187 Meiss, B, 2-179 Slutskln, A.A. 2-326 GOdert, W, 2-204, 2-257 Mall,S. 2·320 Smither, R. 2-340 Goncharuk, E.D, 2-139 Metzler, D.P. 2-253 smith, LInda C.(Ed,) 2·073 Goodstein, LP. 2-269 Metz, p, 2-258 Snlmshchlkova.I,A, 2·227 Goplnath, MA 2·216 Mlscho, W,H, 2·088 Soda. G, 2-102 Mishra, D.S. 2-197 Soroklna, N.G. 2-304 Hahn, U. 2-252 Mittal, R, 2-174 Sollropoulos, MTh. 2-307

126 Int. Classif. l7(I990)No.2 - Classification Literature