Ranking of Coordinate Terms and Hypernyms Using a Hypernym-Hyponym Dictionary
Kosetsu Tsukuda, Hiroaki Ohshima, Katsumi Tanaka · 2014 IEEE/WIC/ACM International Joint Conferences on Web Intelligence (WI) and Intelligent Agent Technologies (IAT) · 2014
In this paper, methods for ranking coordinate terms and hypernyms of a given query according to their appropriateness are proposed. Although previous studies have proposed methods for discovering coordinate terms or hypernyms of a query, they focused on only discovering such terms and evaluating discovered terms based on a binary evaluation: appropriate or inappropriate. Unlike these studies, we rank coordinate terms and hypernyms of a query and evaluate the terms by considering their appropriateness. In the proposed method, a bipartite graph is created based on hypernyms of a query and hyponyms of each hypernym using a hypernym-hyponym dictionary. Subsequently, we apply a HITS-based algorithm to the bipartite graph and rank coordinate terms and hypernyms based on their appropriateness. The experimental results obtained using 50 queries demonstrate that our method could rank appropriate coordinate terms and hypernyms higher than other comparable methods.