WEBODE in a Nutshell

AI Magazine Volume 24 Number 3 (2003) (© AAAI)

Articles WEBODE in a Nutshell

Julio César Arpírez, Oscar Corcho, Mariano Fernández-López, and Asunción Gómez-Pérez

■ WEBODE is a scalable workbench for ontological different languages, are difficult to exchange engineering that eases the design, development, between ontology platforms. The need for and management of ontologies and includes mid- studies of tools interoperability is being ad- dleware services to aid in the integration of ontolo- dressed by the Special Interest Group on Enter- gies into real-world applications. WEBODE presents prise-Standard Ontology Environments of the a framework to integrate new ontology-based tools European Information Society Technologies and services, where developers only worry about thematic network ONTOWEB (IST-2000-29243). the new logic they want to provide on top of the knowledge stored in their ontologies. Taking into account the functions currently provided by available tools and other interest- ing functions that could be added on top of them, we propose the need for an integrated ontological engineering workbench (Corcho et n the last decade, several tools for ontology al. 2002) supporting three main groups of ac- development have appeared: OILED (Bech- tivities (ﬁgure 1): hofer et al. 2001), ONTOEDIT (Sure et al. I First are ontology development and man- 2002), ONTOLINGUA (Farquhar, Fikes, and Rice agement activities that for ontology develop- 1997), ONTOSAURUS (Swartout et al. 1997), PRO- ment include ontology edition, browsing, TÉGÉ-2000 (Noy, Fergerson, and Musen 2000), learning, merge, alignment, translation, and WEBODE (Corcho et al. 2002),1 WEBONTO (Do- evaluation and for ontology management in- mingue 1998), and so on. Comparative studies clude ontology configuration management of some of these tools appeared in Duineveld et and evolution and documentation. 2 al. (1999). Second are ontology middleware services These tools can be used for a great range of that allow the easy use and integration of on- activities in the ontology development tology-based technology into existing and fu- process, such as ontology design, browsing and ture information systems, such as services for implementation, ontology importation, ontol- ontology library administration and access and ogy merge and alignment, and ontology-based ontology upgrading, querying, and metrics. resource annotation. However, only ONTOEDIT, Third are ontology-based application devel- PROTÉGÉ-2000, and WEBODE give integrated opment suites that provide multiple means for support to most of the activities of the ontol- the rapid development and integration of on- ogy development process, and only ONTOEDIT tology-based applications. Ideally, they will be and WEBODE give support to an ontology- the last step toward a real integration of on- building methodology (ONTOKNOWLEDGE [Sure tologies into enterprise information systems. and Studer 2002] and METHONTOLOGY [Fernán- In the context of this framework, WEBODE dez-López et al. 1999], respectively). has been developed as a scalable ontological Moreover, most of the tools have usually engineering workbench that gives support to been built as isolated, independent tools that most of the ontology development and man- are incapable of interoperating. Consequently, agement activities presented in figure 1. WE- ontologies owned by different organizations, BODE also includes middleware services to aid built using different tools and implemented in in the integration of ontologies into real-world

Ontology-Based Applications Ontology Semantic Knowledge Brokers Development Portals Management Suite

Ontology Middleware

Ontology Access Ontology Services Selection Services

Component Based Ontology Library Ontology Metrics Ontologies Upgrading Easy Integration Services Services

Administration Query Services Services

Ontology Ontology Ontology Ontology Ontology Configuration Edition Merge Translation Evaluation Management

Ontology Ontology Ontology Ontology Ontology Learning Browsing Alignment Evolution Documentation Ontology Development and Management

Figure 1. Main Modules of an Ontological Engineering Workbench (adapted from Corcho et al. [2002]).

applications as well as rapid development tools WEBODE Architecture for building ontology-based web portals WEBODE is built completely with JAVA technol- (ODESEW) and ontology-based knowledge ogy, and it is designed according to an n-tier ar- management applications (ODEKM). Finally, chitecture, partitioned into three independent WEBODE was created to provide technological layers: (1) front ends (presentation), (2) middle support to METHONTOLOGY (Fernández-López et (business logic), and (3) back ends (database al. 1999), a methodology for ontology con- management). struction. Nevertheless, this fact does not pre- Current WEBODE front ends, such as the on- vent it from being used based on other tology editor, ODEKM, and ODESEW, are imple- methodologies or no methodological approach mented with HTML and JAVA applets. Because the at all. communication between clients and the mid- The next sections describe the architecture dle layer is carried out with HTTP, front ends can and knowledge model of the WEBODE work- also use web service technologies such as SOAP bench. Then we present the most important and WSDL.3,4 features of the WEBODE ontology editor, and The middle tier is based on an application we describe how to use WEBODE ontologies in server, which makes it easy to create and add ontology-based applications. new services. The application server improves

38 AI MAGAZINE Articles

ODEClean

WebPicker Inference Engine

Import Services Export Services

Ontology Languages Ontology Languages

DAML+OIL OWL DAML+OIL OIL OWL

RDF(S) XCARIN RDF(S) XCARIN FLogic

Other Languages Other Languages OKBC XML Jess Java XML Prolog ODEMerge

ODEAPI

Cache Consistency Axiom

Application Server

Java Virtual Machine

Figure 2. Architecture of WEBODE Services.

WEBODE’s flexibility, scalability, and integra- WEBODE ontologies can be stored in any re- tion with third-party solutions and includes lational database with JAVA database connectiv- basic functions of authentication, log, admin- ity (JDBC) support (WEBODE has been tested istration, thread management, scheduling, with ORACLE and MYSQL). This back end opti- backup, and database access. WEBODE’s cur- mizes database connections by connection rent services are shown in figure 2. The most pooling and provides transparent fault-toler- important service is the WEBODE ontology ac- ance capabilities. In addition, its physical mod- cess APPLICATION PROGRAMMING INTERFACE (API) el is tuned for maximum performance. (ODE API), which allows accessing ontologies in the workbench. As shown in the ﬁgure, the ODE EB service relies on other services for ontology W ODE’s Knowledge Model caching (improving access time to ontology Ontologies in WEBODE are conceptualized components), consistency checking, and ax- with a very expressive knowledge model (Cor- iom checking. A wide range of services for im- cho et al. 2002), which is based on the refer- porting and exporting ontologies are available, ence set of intermediate representations of as we describe in the next section, and some METHONTOLOGY (Fernández-López et al. 1999). other services rely on them. For example, The following ontology components are in- WEBPICKER wraps web resources and converts cluded in WEBODE’s knowledge model: con- them into ontologies and relies on the XML im- cepts and their attributes (both instance and port service. WebODE’s inference engine uses class attributes); concept groups, which repre- ontologies exported in PROLOG, and ODECLEAN sent sets of disjoint concepts; concept tax- uses the WEBODE’s inference engine. Another onomies and disjoint and exhaustive class de- service available is ODEMERGE, which merges compositions; ad hoc binary relations between ontologies. concepts, which can be characterized by rela-

FALL 2003 39 Articles

Clipboard

Browsing Area Edition Area

Figure 3. WEBODE Ontology Editor.

tion properties (symmetry, transitiveness, and WEBODE Ontology Editor so on); constants; formal axioms, expressed in The WEBODE ontology editor is a web applica- first-order logic; rules; and instances of con- tion that allows the collaborative construction cepts and relations. Bibliographic references, of ontologies at the knowledge level. In this synonyms, and acronyms can be attached to section, we present WEBODE’s ontology edi- any of the aforementioned ontology compo- tion functions, its ontology documentation nents. options, its import and export facilities, its The WEBODE knowledge model also allows consistency-checking functions, its inference referring to terms from other ontologies (im- engine, and its ontology evaluation functions. ported terms), which are identified by means of uniform resource identiﬁers (URIs). Ontology Edition WEBODE instances are defined inside in- The user interface of the WEBODE ontology ed- stance sets, which allows creating different in- itor is divided into three main parts: (1) HTML stantiations for the same ontology that are in- forms to edit all ontology components except dependent from each other. For example, if we formal axioms and rules; (2) ONTODESIGNER to have created an ontology about books and li- graphically edit concepts, relations, and con- braries, we can instantiate it in different in- cept taxonomies; and (3) the WEBODE axiom stance sets, one for each library that is using builder (WAB) to edit formal axioms and rules. the ontology. Although the HTML user interface and ONTODE-

40 AI MAGAZINE Articles

Figure 4. Screenshot of ONTODESIGNER.

SIGNER allow creating lightweight ontologies, tive part-of relationships. Ad hoc relations be- WAB allows adding axioms and rules to convert tween concepts can also be created with this them into heavyweight ontologies (Studer, interface. Figure 4 shows a screenshot of ON- Benjamins, and Fensel 1998). TODESIGNER editing an ontology on the domain Figure 3 shows a screenshot of the HTML in- of traveling. terface. It is creating an instance attribute of With ONTODESIGNER, users can create differ- the concept travel. There are three main com- ent views of an ontology to highlight or cus- ponents of this user interface: (1) browsing tomize distinct parts of its concept taxonomy, area, (2) clipboard, and (3) edition area. relations between concepts, and so on. More- The browsing area is used to navigate the over, users can decide at any time whether to whole ontology, including operations to create show or hide different kinds of relations (either new terms and modify or delete existing terms. predeﬁned or ad hoc), in the sense of a graphic The clipboard is used to easily copy and paste prune. This feature helps in the manual evalu- information between forms. ation of the relations contained in an ontol- The edition area presents forms to insert, ogy. delete, and update ontology terms (concept, at- WAB is a graphic editor that allows creating tribute, relation, and so on) and tables with ex- first-order logic axioms and rules. Figure 5 isting ones. For example, figure 3 shows four shows an axiom in WAB that states that “trains attributes of the concept travel that have al- that depart from Europe must arrive in Eu- ready been attached to it—(1) arrival date, (2) rope.” The buttons below the text box help company name, (3) departure date, and (4) sin- with writing logical expressions with quanti- gle fare—and a form to create a new attribute fiers and logical connectives, and the drop- for this concept. down lists below them allow the easy inclusion ONTODESIGNER is a graphic user interface of ontology concepts, their attributes and ad (GUI) used to build concept taxonomies with hoc relations, and constants. the following set of predeﬁned relations: sub- Once an axiom is completely written in the class-of, disjoint decompositions and exhaus- text box, it is parsed and transformed into tive partitions, and transitive and nontransi- Horn clauses through a skolemization process.

FALL 2003 41 Articles

Figure 5. Axiom Edition with WAB.

If it is not possible to transform the axiom into tributes, instances, and ad hoc binary relations. Horn clauses, WAB warns the user. These Horn It also presents a sample result of the HTML doc- clauses are then transformed into PROLOG using umentation service, showing the ontology primitives defined in the open knowledge base concepts, the concept taxonomy (the subclass- connectivity (OKBC) protocol (Chaudri et al. of relationship between concepts is represent- 1997) so that they can be used by the PROLOG ed by indentation), the concept attributes, and inference engine attached to WEBODE. ad hoc binary relations (with their name and Rules are created similarly. They are com- their destination concept). posed of an antecedent and a consequent, and concepts, attributes, relations, and constants Ontology Import and Export Services can easily be included in the rule by means of The WEBODE ontology editor provides transla- drop-down lists. Figure 6 shows WAB creating a tion services for the following ontology lan- rule that states that “trips by ship that depart guages—XCARIN, FLOGIC, RDF(S), OIL, DAML + OIL, from Europe are handled by the company and OWL—and translation services from XCARIN, CostaCruises.” As with formal axioms, rules are RDF(S), DAML + OIL, and OWL. Taking into account also transformed into Horn clauses and then to that the WebODE knowledge model is very ex- OKBC-based PROLOG rules. pressive, we are able to provide high-quality translations that preserve most of the original Ontology Documentation information contained in the ontology and WEBODE ontologies are automatically docu- take advantage of most of the modeling char- mented in different formats, such as the al- acteristics of the target languages. ready mentioned METHONTOLOGY’s intermediate WEBODE ontologies can also be exported and representations, HTML, and XML. The whole on- imported from XML, following a well-defined tology, or parts of it (specific views or instance document-type definition (DTD) that uses the sets), can be selected for this documentation. same knowledge representation vocabulary Figure 7 presents the concept dictionary of that is used to express its knowledge model. an ontology in the traveling domain, includ- It can also export ontologies into other lan- ing all its concepts, their class and instance at- guages that are not specifically designed for im-

42 AI MAGAZINE Articles

Figure 6. Rule Edition with WAB.

plementing ontologies, such as PROLOG, JAVA, 2002) by means of the ODECLEAN service (Fer- and JESS. nández-López and Gómez-Pérez 2002). This evaluation service uses the WEBODE inference Ontology Consistency Checking engine and can be switched on and off so that When building ontologies with the WEBODE WEBODE users can decide whether to use it to ontology editor, their consistency is checked. clean their concept taxonomies according to The ontology editor checks type constraints, the notions of rigidity, identity, and unity that numeric value constraints, cardinality con- are used by this method. straints, and concept taxonomy consistency Figure 8 shows ONTODESIGNER cleaning the (that is, common instances of disjoint con- concept taxonomy of a travel ontology with cepts, loops in the concept taxonomy, and so ODECLEAN. The symbols that appear in the on). These consistency-checking functions are lower part of each concept represent that a invoked either from the HTML interface or ON- concept is rigid (+R), is nonrigid (-R), is anti- TODESIGNER. rigid (~R), has unity (+U), and so on. The symbols given in parens are typically used in the Built-In Inference Engine method. After applying these metaproperties The WEBODE ontology editor includes a built- to the concepts, the user can evaluate the con- in OKBC-based inference engine, which is cept taxonomy, and the errors in the taxono- based on a subset of the primitives identiﬁed in my appear in a separate window and are high- that protocol (Chaudhri et al. 1997). This infer- lighted in the graph. ence engine is implemented in CIAO PROLOG and can be used to query information about the How to Use WEBODE Ontologies ontologies and detect more types of inconsis- tencies with the axioms created with WAB.5 Ontologies built with the WEBODE ontology editor or imported into the workbench can be Ontology Evaluation used by ontology-based applications in many The WEBODE ontology editor gives support to different ways: the ONTOCLEAN method (Guarino and Welty Ontologies can be accessed and modified

FALL 2003 43 Articles

Figure 7. Different Types of Documentation of the WEBODE Ontology Editor.

with the WEBODE ontology access API (ODE er Prolog programs, and integrate them into API in figure 2). Ontology-based applications ontology-based applications that use these Pro- can be built as services on top of the same ap- log programs. plication server as WEBODE or can access WEB- Ontologies can be exported to JAVA. Each on- ODE ontologies remotely with remote method tology concept is transformed into a JAVA class, invocation (RMI) or web services. each attribute is transformed into a variable, Ontologies can be exported to XML so that each ad hoc relation is transformed into a class the ontology-based application accesses and association, and so on. This JAVA code can be modifies the XML file obtained. If the ontologies compiled and used in other JAVA applications. are modified, they can later be imported into the WEBODE workbench if needed. Ontologies can be exported to several ontol- Conclusions ogy languages (XCARIN, FLOGIC, RDF(S), OIL, DAML In this article, we presented the WEBODE onto- + OIL, OWL) and other more general languages logical engineering workbench and its main (JESS), as presented in the previous section. Spe- contributions to the ontology area, which are cific APIs and inference engines developed by summarized as follows: third parties for these languages can be used to WEBODE gives integrated technological sup- access, modify, and reason with these ontolo- port to several activities of the ontology devel- gies. For example, if an ontology is trans- opment process, usually supported by other in- formed into RDF(S), there are several parsers and dependent tools that do not interoperate. inference engines that could be used, such as WEBODE gives support to ontology edition, JENA, the ICS-FORTH RDFSUITE, SESAME, and RED- ontology translation, ontology merge, ontol- LAND. With DAML + OIL ontologies, we can use JE- ogy evaluation, and ontology documentation. NA, FACT, RACER, and so on. WEBODE ontology developers do not need Ontologies can be exported to Prolog and to worry about building ontologies directly in used by the WebODE’s OKBC-based Prolog in- an ontology language because WEBODE on- ference engine. We can also use them with oth- tologies are automatically translated into many

44 AI MAGAZINE Articles

Figure 8. Ontology Evaluation with ODECLEAN. implementation languages. Moreover, first-or- In the Spanish Comisión Interministerial de der logic axioms and rules can easily be creat- Cienca y Tecnolgía (CICYT) project on ed with WAB. Methodology for Knowledge Management WEBODE has been built to support an ontol- (TIC-980741), we built ontologies that mod- ogy development methodology—METHONTOL- eled institutions and developed the core ser- OGY—but it does not prevent WEBODE from be- vices of the workbench as well as the ODEKM ap- ing used in the context of other methodologies plication. for ontology development or without any spe- In the Spanish CICYT project CONTENTWEB cific methodological approach. It also provides (TIC2001-2745), we created WEBPICKER to per- a service that gives support to the ONTOCLEAN form ontology learning from electronic-com- method for ontology evaluation. merce (e-commerce) standards for product and WEBODE ontologies can be used in ontol- service classification (UNSPSC, ROSETTANET, and E- ogy-based applications in many different ways: CL@SS). We also built ontologies in the domain (1) using its JAVA API directly, (2) using with of leisure activities. RMI or web services from remote applications, In the UPM project Environment Ontology (3) using XML files, (4) exporting and importing (UPM-AM-9819), we built ontologies in the do- ontologies in different ontology languages main of chemistry—elements and environ- from and into the workbench, (5) exporting mental ions—and integrated them with exist- ontologies to general languages such as PROLOG ing ontologies, such as the ONTOLINGUA or JESS, and (6) exporting ontologies to JAVA. ontology STANDARD UNITS. The WEBODE workbench was built and test- In MRO (ontologies for cataloging business ed in the context of the following projects, services), we merged heterogeneous electronic where we also developed several ontologies: catalogs in the domain of office furniture.

FALL 2003 45 Articles

tools, and languages for building ontologies as well as ontology-based applications. In the European project Esperonto (IST- Useful URLs 2001-34373), we built ODESEW, which is an ontology-based web portal that automatically creates an intranet and an extranet from WE- BODE ontologies. In this project, ODESEW is Ontology Tools being used with ontologies that model a re- OILED: oiled.man.ac.uk/ search and development project, including organizations and people participating in the ONTOEDIT: www.ontoprise.de/products/ontoedit project, documentation, work packages, and so ONTOLINGUA: ontolingua.stanford.edu on. ONTOSAURUS: www.isi.edu/isd/ontosaurus.html Finally, and also in the context of the Euro- pean project Esperonto (IST-2001-34373), we PROTEGE-2000: protege.stanford.edu/ are creating more middleware services and WEBODE: webode.dia.fi.upm.es/ more ontology development and management WEBONTO: kmi.open.ac.uk/projects/webonto/ services, such as ontology configuration management and evolution capabilities, ontology upgrading, semantic annotation services, on- WEBODE Applications tology learning, and more integrated ontology ODESEW: webode.dia.fi.upm.es/sew/index.html reasoning services. We are laying the founda- tion for a more complete implementation of ONTOROADMAP: webode.dia.fi.upm.es/ontoweb/wp1/ OntoRoadMap/index.html the workbench presented in the first section.

ODEKM: forgy.dia.ﬁ.upm.es/km/login.html Acknowledgments

MKBEEM: www.mkbeem.com/ This work is supported by grant AP2002-3828 funded by Ministerio de Educación y Cultura ONTOWEB Special Interest Group on Enterprise-Standard and by the projects CONTENTWEB (TIC-2001- Ontology Environments: delicias.dia.ﬁ.upm.es/ontoweb/ sig-tools/index.html 2745), MKBEEM (IST-1999-10589), ONTOWEB (IST-2000-29243), and Esperonto (IST-2001- 34373). This workbench could not have been Ontology Markup Languages created without the help of R. de Diego, R. RDF(S): www.w3.org/RDF/ González, M. Lama, A. López, V. López, J. P. OIL: www.ontoknowledge.org/oil/ Pérez, E. Raya, and O. Vicente in the implementation and tests of WEBODE services. DAML + OIL: www.daml.org/language/

OWL: www.w3.org/TR/owl-ref/ Notes 1. webode.dia.ﬁ.upm.es/. 2. Ontology Building: A Survey of Editing Tools, www.xml.com/pub/a/2002/11/06/ontologies.html. 3. www.w3.org/TR/SOAP/. 4. www.w3.org/TR/wsdl. 5. www.clip.dia.fi.upm.es/Software/Ciao/index.- html. The WEBODE workbench has been used to develop ontologies and ontology-based appli- References cations in the following projects: Bechhofer, S.; Horrocks, I.; Goble, C.; and Stevens, R. In the European project Multilingual Knowl- 2001. OILED: A Reasonable Ontology Editor for the edge-Based European Electronic Marketplace Semantic Web. In Joint German/Austrian Conference (MKBEEM) (IST-1999-10589), we built and on Artificial Intelligence (KI’01), eds. F. Baader, G. Brewka, and T. Eiter, 396–408. Lecture Notes in Arti- reengineered business-to-business (B2B) and ficial Intelligence 2174. Vienna, Austria: Springer- business-to-consumer (B2C) ontologies. These Verlag. ontologies have been used in an e-commerce Chaudhri, V. K.; Farquhar, A.; Fikes, R.; Karp, P. D.; platform using the XCARIN translator. and Rice, J. P. 1997. The Generic Frame Protocol 2.0. In the European thematic network ONTOWEB Technical Report, Knowledge Systems Laboratory, (IST-2000-29243), we built the ONTOROADMAP Stanford University. application, which uses the WEBODE API to ac- Corcho, O.; Fernández-López, M.; Gómez-Pérez, A.; cess the ontologies and allows users to register, and Vicente, O. 2002. WEBODE: An Integrated Work- browse, and search ontologies, methodologies, bench for Ontology Representation, Reasoning, and

46 AI MAGAZINE Articles

Exchange. Paper presented at the Thirteenth Interna- Julio César Arpírez has a BA in tional Conference on Knowledge Engineering and computer science (2000) from the Knowledge Management (EKAW2002), 1–4 October, Universidad Politécnica de Madrid Sigüenza, Spain. (UPM). He is currently a Ph.D. stu- Domingue, J. 1998. TADZEBAO and WEBONTO: Dis- dent in AI at UPM. He is currently cussing, Browsing, and Editing Ontologies on the working for a startup company, fo- Web. Paper presented at the Eleventh International cusing on the 2.5G and 3G markets. Workshop on Knowledge Acquisition, Modeling, His main interests include AI and and Management (KAW’98), 18–23 April, Banff, enterprise and telecommunication systems. His e- Canada. mail address is [email protected].fi.upm.es. Duineveld, A.; Stoter, R.; Weiden, M. R.; Kenepa, B.; Oscar Corcho has a BA in computer and Benjamins, V. R. 1999. Wonder Tools? A Com- science (2000) and an MSc in soft- parative Study of Ontological Engineering Tools. Pa- ware engineering (2001) from the per presented at the Twelfth Workshop on Knowl- Universidad Politécnica de Madrid (UPM). He is currently a Ph.D. stu- edge Acquisition, Modeling, and Management dent in AI at Universidad Politécni- (KAW99), 16–21 October, Banff, Canada. ca de Madrid (UPM). He received Farquhar, A.; Fikes, R.; and Rice, J. 1997. The ONTOLIN- the third Spanish award for com- GUA Server: A Tool for Collaborative Ontology Con- puter science from the Spanish government in 2001. struction. International Journal of Human Computer His research activities include ontology languages, Studies 46(6): 707–727. ontology development tools, the ontology transla- Fernández-López, M., and Gómez-Pérez, A. 2002. tion problem, and the semantic web. He is participat- The Integration of ONTOCLEAN in WEBODE. Paper ing in the ONTOWEB thematic network (IST-2000- presented at the EKAW2002 Workshop on Evalua- 25056) and the IST Esperonto project (IST-2001- tion of Ontology-Based Tools (EON2002), 30 Septem- 34373). His e-mail address is ocorcho@fi.upm.es. ber, Sigüenza, Spain. Mariano Fernández-López has a Fernández-López, M.; Gómez-Pérez, A.; Pazos, J.; and BA in computer science (1996), an Pazos, A. 1999. Building a Chemical Ontology Using MSc in knowledge engineering METHONTOLOGY and the Ontology Design Environ- (2000), an MSc in software engi- ment. IEEE Intelligent Systems and Their Applications neering (2000), and a Ph.D. in com- 4(1): 37–45. puter science (2001), all from the Guarino, N., and Welty, C. Evaluating Ontological Universidad Politécnica de Madrid Decisions with ONTOCLEAN. Communications of the (UPM). He is an associate professor ACM 45(2): 61–65. in the Computer Science School at UPM and an associate professor at the Technical School of Fun- Noy, N. F.; Fergerson, R. W.; and Musen, M. A. 2000. dación Universitaria San Pablo CEU. His research ac- The Knowledge Model of PROTEGE-2000: Combining tivities include ontology development and merging Interoperability and Flexibility. Paper presented at methodologies and ontology tools. His e-mail ad- the Twelfth International Conference in Knowledge dress is mfernandez@fi.upm.es. Engineering and Knowledge Management (EKAW- 00), 2–6 October, Juan-Les-Pins, France. Asunción Gómez-Pérez has been the director of the Ontology Group Studer, R.; Benjamins, V. R.; and Fensel, D. 1998. at the Universidad Politécnica de Knowledge Engineering: Principles and Methods. Madrid (UPM) since 1995. She has a IEEE Transactions on Data and Knowledge Engineering BA in computer science (1990), an 25(1–2): 161–197. MSc in knowledge engineering Sure, Y., and Studer, R. 2002. On-to-Knowledge (1991), and a PhD in computer sci- Methodology. In On-to-Knowledge: Semantic Web–En- ence (1993) from the Universidad abled Knowledge Management, eds. J. Davies, D. Politécnica de Madrid (UPM). She also has an MBA Fensel, and F. van Harmelen, 33–46. New York: Wi- (1994) from Universidad Pontificia de Comillas. She ley. is currently a research adviser at the AI lab at UPM. Sure, Y.; Erdmann, M.; Angele, J.; Staab, S.; Studer, R.; Her current research activities include interoperabil- and Wenke, D. 2002. ONTOEDIT: Collaborative Ontol- ity between different kinds of ontology development ogy Engineering for the Semantic Web. In Proceedings tools, methodologies and tools for building and of the First International Semantic Web Conference merging ontologies, ontological reengineering, on- (ISWC’02), eds. I. Horrocks and J. A. Hendler, tology evaluation, and ontology evolution as well as 221–235. Lecture Notes in Computer Science 2342. uses of ontologies in applications related to the se- Berlin, Germany: Springer-Verlag. mantic web, e-commerce, and knowledge management. Her e-mail address is asun@fi.upm.es. Swartout, B.; Ramesh, P.; Knight, K.; and Russ, T. 1997. Toward Distributed Use of Large-Scale Ontolo- gies. Paper presented at the Spring Symposium on Ontological Engineering, 24–26 March, Stanford, California.

FALL 2003 47 Call for Proposals

Call for Proposals AAAI-04 Tutorial Forum Nineteenth National Conference on Artiﬁcial Intelligence July 25-29 San Jose, California Sponsored by the American Association for Artiﬁcial Intelligence

he AAAI-04 Program Committee invites proposals for for continued innovation in the forum's program that incor- the Tutorial Forum of the Nineteenth National Confer- porates novel and under-represented topic areas. Tence on Artificial Intelligence (AAAI-04). The Tutorial Forum will be held July 25–29, 2004 in San Jose, California. Submission Requirements Anyone interested in presenting a tutorial at AAAI-04 should We need two kinds of information in the proposals: informa- submit a proposal to Matthew Stone, 2004 Tutorial Forum tion that will be used for selecting proposals and information Chair, at the address below. that will appear in the tutorial description brochure. The pro- What Is the Tutorial Forum? posal should provide sufficient information to evaluate the quality of the technical content being taught, the quality of The Tutorial Forum provides an opportunity for junior and se- the educational material being used, and the speakers' skill at nior researchers to spend two days each year freely exploring presenting this material. Each proposal should include at least exciting advances in disciplines outside their normal focus. We believe this type of forum is essential for the cross fertil- the following: ization, cohesiveness, and vitality of the AI field. We all have • Goal of the tutorial: Who is the target audience? What will a lot to learn from each other; the Tutorial Forum promotes the audience walk away with? What makes the topic inno- the continuing education of each member of the AAAI. vative? • Content: Detailed outline and list of additional materials, Topics augmented with samples, such as past tutorial slides and AAAI is interested in proposals for advanced tutorials at the survey articles, whenever possible. Be as complete as possi- leading edge of AI. We especially encourage tutorials taught ble. by a strong team of two established researchers, providing a • Tutorial description: A short paragraph summarizing the tu- balanced perspective on a topic of broad potential interest across the AI community. We are interested in tutorials that torial outline. explore a specific research area in depth by summarizing its • Prerequisite knowledge: What knowledge is assumed. recent technical and methodological advances, educating Please also submit the following information about the participants about its important outstanding challenges and team of presenters: name, mailing address, phone number, e- opportunities, and exploring its general ramifications for AI mail address; background in the tutorial area, including a list practice, including training as well as research. Our goal is to of publications and/or presentations; any available examples present a diverse program that includes core areas of AI, new of work in the area (ideally, a published tutorial-level article or techniques from allied disciplines that can inform research presentation materials on the subject); evidence of teaching within AI, and conversely emerging applications of AI tech- experience (courses taught or references); and evidence of niques to new areas. Previous years' tutorial programs provide an indication of the scope and variety of possible topics. The scholarship in AI or computer science. list is not exclusive; indeed, we are expressly interested in top- Submission Deadline ics that we would not have imagined to mention. Finally, note that we very much welcome proposals for educational Proposals must be received by October 15, 2003. Decisions approaches that go beyond the traditional format of four- about the tutorial program will be made by December 1, hour tutorials, exploiting the flexibility that the one-fee pro- 2003. Speakers should be prepared to submit completed gram offers. course materials by May 25, 2004. AAAI-02’s forum included tutorials on language modeling; Please e-mail proposal material to the tutorial chair at the AI in space; greedy on-line planning; qualitative spatial and following address. Hard copy submissions will also be accept- temporal reasoning; practical approaches to handling uncer- ed: tainty in planning and scheduling; collaborative multi-agent systems; practical machine learning for software engineering; • Matthew Stone information integration on the web; AI techniques for per- Department of Computer Science — Rutgers University sonalized recommendation; algorithms for combinatorial 110 Frelinghuysen Road auctions and exchanges; phase transitions and structure in Piscataway NJ 08854-8019 combinatorial problems; and rational action in autonomous Telephone: 732 445 3546 / Fax: 732 445 1494 agents. This list serves merely as an example. We are looking E-mail: [email protected]

48 AI MAGAZINE