In this study, we have constructed an alignment table between 'Word List by Semantic Principles (revised and enlarged edition)' (hereafter WLSP) and UniDic to develop large-scale Japanese corpora which is comprehensively annotated with systematic word senses. WLSP is an extensive contemporary Japanese thesaurus with systematic semantic categories. UniDic is a vast lexicon used for Japanese morphological analysis and is utilized in the development of large-scale Japanese corpora. The alignment table defines n-to-n same word relations between 64,759 WLSP entries and 50,795 UniDic lexemes. These relations were manually verified based on scripts, readings, and classes of the WLSP entries and the UniDic lexemes. The development of word-sense annotated Japanese corpora has commenced with the use of the table. A Japanese morphological analysis tool to annotate word-sense was also developed with the table. Meanwhile, for a full-scale development of word-sense annotated Japanese corpora, it is necessary that problems, such as enlargement of the table and identification of word senses in corpora are effectively dealt with.
「分類語彙表番号－UniDic語彙素番号対応表」の構築
papers1804.pdf
(723.55KB)
概要
