Question Matching Based on Domain Ontology and Description Logic

Abstract

Question matching is a key stone in web intelligent question answering system. A novel question matching algorithm based on ontology and description logic is presented in the paper. Domain knowledge represented by ontology is very important in question matching. Concepts and relations between concepts are extracted from both standard and user questions, then the concepts and relations in standard questions are translated to description logic (DL) rules, while concepts and relations in user question are translate to predicate. User questions and standard questions are matched based on description logic and matching ranking technique. The formal and semantic ability of the question matching method derived from ontology and description logic. The experimental results show the question matching algorithm is superior to methods based on word retrieve technique in precision and recall.