Abstract
Sentence similarity computing plays an important role in the Question Answering (QA) System. Because there are many question expressions for one meaning, we present a new approach to match question based on Fuzzy set. In this paper, we establish a library of standard questions. Each standard question is relative with a series of Keywords. The main focus of this paper lies with matching of standard questions and questions asked by users. An experimental system based on the proposed method has been built, and the results of our experiments shows the proposed method is effective for question matching.