Learning Ontologies
Learning Web Service domain ontologies
WonderWeb pioneered the use of ontology learning tools in the context of Web services by investigating and (partially) answering a set of fundamental questions. The most important observation was that the textual sources have a low grammatical quality and that they use language in a specific way (i.e., they employ a sublanguage). These characteristics constrain the choice of approaches that can be used. Also, the concepts that should be extracted, should reflect both static (domain concepts) and procedural (functionalities that Web services offer) aspects of the domain.
This situation is different from current ontology learning where only static knowledge is extracted. Therefore existing methods have to be adapted. In addition, important limitations still exist which will be tackled in TAO. In particular, the methods only extract simple information about services (their parameters and functionality), while much more can be learned form these descriptions. Such information can be extracted by extending the current methods or by considering other sources as well (e.g., recent experiments have shown that WSDL descriptions also contain valuable and often more precise and detailed information than the textual documentations). Another challenge is combining the knowledge extracted from different sources.
Related Publications
Mladenic, Dunja, Grobelnik, Marko, Fortuna, Blaz, "What Semantic Web researchers need to know about Machine Learning?", SWAP 2007, December 18-20, 2007. Video.
Mladenic, Dunja, Grobelnik, Marko, Fortuna, Blaz, "What Semantic Web researchers need to know about Machine Learning?", In Proceedings of the International Semantic Web Conference (ISWC 2007), Busan, Korea, November 11-15, 2007.
Grcar, Miha, Grobelnik, Marko and Mladenic, Dunja. "Using Text Mining and Link Analysis for Software Mining". ECML/PKDD, September, 2007.
, Presentation.
Mladenić, Dunja, Grobelnik, Marko. "Evaluation of semi-automatic ontology generation in real-world setting". In Proceedings of the 29th International Conference on Information Technology Interfaces, Cavtat/Dubrovnik, Croatia, June 25-28, 2007.
.
Ilijašić Mišić, Ivana, Kovačič, Božidar, Mohorić, Tamara, Mladenić, Dunja, Fortuna, Blaž, Grobelnik, Marko. "User study of ontology generation tool". In Proceedings of the 29th International Conference on Information Technology Interfaces, Cavtat/Dubrovnik, Croatia, June 25-28, 2007.
.
Fortuna, Blaz, Grobelnik, Marko and Mladenic, Dunja. "OntoGen: Semi-automatic Ontology Editor". HCI International 2007, Beijing, July 2007.
.
Fortuna, B. and Grobelnik, M. and Mladenic, D. "Semi-automatic Ontology Construction System". AAAI 2007, Vancouver, July 2007.
.
Related Deliverables
D2.1 - Data source analysis and method selection
D2.2.1 - Ontology learning services library - v1
D2.2.2 - Ontology learning services library - v2
D2.2 - Software - Ontology Learning Software
D2.3 - Ontology learning: evaluation report
D5.1 - Evaluation of existing SWS tools and infrastructure
D5.2 - Architecture and integration requirements and specification
D5.3.1 - TAO Suite implementation and user manual -v1
D5.3.2 - TAO Suite implementation and user manual -v2
D5.4.1 - TAO Suite best practice documentation - v1
D5.4.2 - TAO Suite best practice documentation - v2
Related Materials
Software: Latino/OntoGen, OntoSight
Demos: Latino/OntoGen, OntoSight, OL for Aeronautics
Further Materials
If you would like to receive news about the TAO project, you can subscribe here.

