摘要： 提出一种利用概念图计算问句相似度的方法，并在此基础上实现基于概念图的中文FAQ问答系统，在该系统中采用概念图的形式表示用户问句及在FAQ库中找到的候选问句集中的问句，通过改进的概念图语义相似度计算问句相似度，在候选问句集中找到相似的问句并将答案返回给用户。该系统能够自动更新和维护FAQ库。实验结果表明，与基于关键词的句子相似度相比，基于语义的句子相似度提高了问题匹配的准确率。

关键词: 概念图, 相似度, 常问问题集, 问答系统

Abstract: A method of computation sentence similarity based on concept graph is proposed and the Chinese Frequently Asked Question(FAQ) set system based on concept graph is implemented. The concept graph is used to express users query and candidate questions. The semantic similarities of sentences are computed between users query and candidate questions by semantic similarity based on concept graph. The answer in correspondence with the most similar query is returned to the user. This system can also automatically update and maintain FAQ. Experimental results show the new computing method gets better performance than the keywords-based approach.

Key words: concept graph, similarity, Frequently Asked Question(FAQ) set, question-answering system

中图分类号: 

• N945