Semantic relation extraction for automatically building domain ontology using a link grammar
Junho Choi, Chang Choi, Dongjin Choi, Jin-Gwang Koh, Pankoo Kim · 2011
Ontology which is the fundamental for semantic web is getting important while the semantic information processing is developed. To build and extend ontologies, much of research has been proposed such as semi-automatic and full-automatic methods through analyzing raw text documents. However, the methods based on document set in specific domain have a limitation that the ontology depends on the domain. Therefore, this research has used Wikipedia document set for extension of generalized ontology. The Wikipedia contains unrestricted subjects and one document is filled with one subject in detail. Moreover, since the content is written by domain specialist, we can say that Wikipedia provides trustworthy contents. This research which deals with ontology extension, it extracts important sentences from Wikipedia documents and the sentences are structured by Link Grammar. Finally, the ontology extension is accomplished through conceptualization step which grasps subjects, objects, and predicates from the structured contents. In the performance evaluation which compared our method to the other method using Context-Free Grammar showed better accuracy more than 12% points.