基于Web的中文开放式问题回答系统

A Web-based Chinese Open-domain Question Answering System

互联网正逐渐成为重要的信息资源,然而大多数搜索引擎不能处理自然语言提出的问题。基于互联网的中文问题回答系统由问题处理、信息检索、答案抽取和答案判断组成,利用命名实体识别、语义依存关系和案例规则模板实现答案抽取。实验表明:命名实体识别、语义依存关系和案例规则模板能有效地实现答案抽取,获得较高正确率。

The Internet is increasingly being used as a source of reference information. Most popular search engines, however, are not designed for answering natural language questions. A Web-based Chinese question answering system is made of four parts： question processing, information retrieval, answer extraction and answer justification. It utilizes named entity recognition, semantic dependency relations and case-based rule to realize answer extraction. Experiments show that named entity recognition, semantic dependency relations and case-based rule perform very well in answer extraction.