Research on method of Chinese question similarity calculation in restricted domain
Chinese question similarity calculation is one of the key factors in Chinese question answering system. This paper analyzes the current method of similarity calculation and some of the deficiencies, proposed the similarity calculation method of question which combining syntactic, semantic and domain knowledge base in the restricted domain .Picking up syntactic dependences tree of a question by syntactic parsing, by means of semantic knowledge base HowNet to do question lexical semantic parsing, and resolving domain concept relativity by construct domain knowledge base, and using the concept relationship and concept attribution relationship in domain knowledge base, and unifing the characters of domain question to realize question similarity calculation. The result of experiments shows that this method has obtained a better effect.
restricted domain question similarity the keywords feature the syntactic feature domain knowledge base 1
ZHANG Cheng YU Zhengtao DENG Jinhui LIU Gengyan HUANG Yuejuan
The School of Information Engineering and Automation, Kunming University of Science and Technology, The School of Information Engineering and Automation, Kunming University of Science and Technology,
国际会议
北京
英文
2007-08-05（万方平台首次上网日期，不代表论文的发表时间）