基于句型系统的中文问题理解研究

Research on Chinese Question Analysis Based on Sentence Type System

提出了一种基于向量空间模型的句型识别方法,该方法通过基于句型模式的问句句法树规约,得到句子关于句型模式的句型结构,再计算句型结构与句型模式之间的相似度来实现问句的句型识别.并在汉语疑问句句型系统的基础上,通过句型识别实现了对问题的分类和问题理解.测试结果表明,该方法提高了问题理解的准确度.

A method of STR （Sentence type recognition） based on VSM （Vector space model） is put forward like this： analyzing the phase syntactic tree of a question sentence based on the sentence type pattern to get the sentence type structure about the sentence type pattern of the question, and computing the similarity between the sentence type structure and the sentence type patterns to achieve the question＇s sentence type recognition. The experimental results show that the question analysis implemented by this way increases the precision of question classification.