Extracting hyponymic relations from Chinese free corpus
Lei Liu, Cungen Cao, Haitao Wang · 2006
Abstract:- Research on hyponymy acquisition is a basic and crucial problem in knowledge acquisition from text. In this paper we present a method of hyponymic relation acquisition and verification based on Chinese lexico-syntactic patterns. Firstly, we make use of removable lexicons and sentence patterns that have been semi-automatically obtained to analyze Chinese-isa patterns. Then we use an algorithm that combines outside layer removal and inside layer gathering to acquire hyponymic concept. In the final phase, we combine self features and context features together for the verification of hyponymy. Experimental results show that the method is adequate of extracting hyponymy from Chinese free text.