Discover hierarchical lexical hyponymy relation from large-scale concept set

Abstract

Lexical hyponymy relation is a kind of hyponymy that can be directly inferred from the lexical compositions of concepts, and of great importance in ontology learning. However, there is a key problem that the lexical hyponymy is so commonsensible that it cannot be discovered by any existing acquisition methods. In this paper, we propose a novel approach to semi-automatically discover hierarchical lexical hyponymy relations from a large-scale concept set, instead of analyzing lexical structures of concepts. Firstly we design Common Suffix Tree (CST) to cluster lexical concepts. We extract class concepts candidates from CST by statistic-base rules that we investigated, and then use a Google-based classifier to verify them. Finally, we extract lexical hyponymy relation candidates from CST and judge them after a prefix clustering process. Experimental result shows us that our approach can correctly discover most lexical hyponymy relations in a given large-scale concept set.