Abstract: Answer extraction is the key technology of the automatic question answering system.According to analysis of question-answer pairs，there is not only semantic similarity，but also syntactic relevance between them.This paper proposes a new method based on bicharacteristic fusion of answer pattern and semantic features.Using the weight to describe the contribution of each feature，a better MRR can be made for each question category.Compared to other methods，this method can fully describe the answer’s features，and then more accurate result can be achieved.

Key words: answer extraction, question answering, feature fusion, sentence similarity, pattern matching

摘要： 答案抽取是问答系统的关键技术。根据对问题-答案的分析可知，它们之间不仅存在语义上的相似性，而且还有句法上的关联性。提出了一种基于问题-答案之间句法模式和句子语义二元特征融合的答案抽取方法。通过对不同的特征赋予不同的权值来调节各特征对答案抽取准确率计算的贡献。与其他方法相比，描述答案的信息更加全面，答案抽取更加准确。

关键词: 答案抽取, 问答系统, 特征融合, 句义相似, 模式匹配