Correction and Extension of WordNet 1.7 for Knowledge-based Applications
Philippe A. Martin · 2003
This article presents the transformation of the noun-related part of WordNet (108,000 nouns, 74,500 categories representing their meanings, and 95,000 semantic links between them) into a genuine “lexical ontology”, usable for knowledge representation, sharing and retrieval on the Web. To do so, (i) I generated intuitive identifiers for all the categories, (ii) introduced 353 lexical corrections (mainly by adding category names), (iii) extracted 6211 individuals (0th-order categories) to differentiate the WordNet specialization links into subtype and instance links, (iv) added a toplevel ontology of about 150 concept types and 150 primitive relation types to permit normalized knowledge statements and some semantic checks on them and the ontology, (v) removed or modified 306 links between categories to repair inconsistencies or avoid redundancies, and (vi) added 159 links between categories, 50 prototypical schemas of relations from common categories, and about 1300 semantic annotations. These corrections are documented at http://www.webkb.org/doc/wn/ and the ontology is downloadable in DAML/RDF or more concise and expressive formats. Web users can also search and extend the ontology via the WebKB-2 knowledge server at http://www.webkb.org. Hence, the above numbers may increase as the ontology improves. This article illustrates these modifications and their rationales.