摘要： 根据用户提交的查询请求，利用统计语言模型计算查询请求和问句之间的相似度，确定用户查询词所代表的信息需求，由此从海量数据中检索出可以满足该信息需求的问答对，并使用答案质量评估模型对其进行评估。实验结果表明，该问题检索模型可以根据用户请求提供具有较高质量的问答对答案。

关键词: 问答对数据, 统计语言模型, 答案质量评估, 问题检索, 语义相似度

Abstract: According to the search requests submitted by users, this paper adopts the statistical language model to measure the similarity between request and question, so that the information users need can be determined. It retrieves appropriate question-answer data which can meet the information need from the large-scale question-answer data, and ranks the answers according to their quality evaluations. Experimental results show that the model can provide answers with high-quality according to the users search requests.

Key words: question-answer data, statistical language model, answer quality evaluation, question retrieval, semantic similarity

中图分类号: 

• TP391