Semantic Classification of Automatically Acquired Nouns using Lexico-Syntactic Clues

In this paper, we present a two-stage approach to acquire Japanese unknown morphemes from text with full POS tags assigned to them. We first acquire unknown morphemes only making a morphologylevel distinction, and then apply semantic classification to acquired nouns. One advantage of this approach is that, at the second stage, we can exploit syntactic clues in addition to morphological ones because as a result of the first stage acquisition, we can rely on automatic parsing. Japanese semantic classification poses an interesting challenge: proper nouns need to be distinguished from common nouns. It is because Japanese has no orthographic distinction between common and proper nouns and no apparent morphosyntactic distinction between them. We explore lexico-syntactic clues that are extracted from automatically parsed text and investigate their effects.