DFKI-LT - Web-based Ontology Learning with ISOLDE

Nicolas Weber, Paul Buitelaar
Web-based Ontology Learning with ISOLDE
2 Proc. of the Workshop on Web Content Mining with Human Language at the International Semantic Web Conference, Athens GA, USA, 2006
 
The last few years saw a continuing increase in the availability of open source web-based information resources such as Wikipedia and similar initiatives. As this kind of information is freely available, has some pre-defined format, covers many topic areas and is of an acceptable quality it is an ideal resource for bootstrapping or extending domain ontologies. In this paper we show how web resources such as Wikipedia, Wiktionary and a German online dictionary (DWDS) can be used in combination with a domain corpus and a general purpose named-entity tagger to derive a domain ontology. The ISOLDE (Information System for Ontology Learning and Domain Exploration) system we describe generates a domain ontology by extracting class candidates from the linguistic context of a given set of ontology instances and by deriving further knowledge on these class candidates from available web resources.
 
Files: BibTeX, ISWC06.WebContentMining.pdf