Web问答系统中问句理解的研究

对问答系统中的问句理解技术进行了深入研究,提出了对问句信息进行深层挖掘形成问句表征.对问句进行分词、去停用词等预处理;结合FAQ库和网络对问句进行关键词扩展,以网络为语料库,利用N元语法模型对问句中的新词进行识别,利用规则的方法对问句进行分类;利用原始关键词+扩展词+新词+类别的形式对问句的信息进行表征.基于上述理论实现一个问答系统并进行了验证,实验表明:文中的问句理解方法能有效改善系统的性能.

Study on Question Understanding of Web-based Question-answering System

The question understanding technology of question-answering system is studied,and the question information is mined deeply in the question analysis module.Firstly,the question is processed with segmentation and stop words removing.Then,the key word expansion method that combines FAQ and network is proposed,the new words are identified by using N-gram grammar and with the network as corpus,and the questions are classified with the regulations method.Finally,a question is represented by the original keywords and the expanding words and new words and category.A system is designed based on these theories to verify them,experiment results shows: this question understanding technology can improve system performance effectively.