摘要
为了提高问答系统对问句理解的准确率,以概念层次网络理论结合传统计算语言学为思路,提出了适用于限定领域中问句分析模型,并根据限定领域的知识特点,设计了新的问句分类方法。在此问句分类方法的基础上,改进了基于多元贝努里模型的贝叶斯分类算法。在以实际教学过程中所收集的真实问句为问题集和训练集的测试中,取得了较好的实践效果。
A novel closed-domain oriented question analysis module based on hierarchical network of concepts and traditional computational linguistics is proposed to enhance the rate of accuracy of question interpretation of a question answering system, A new question catalog is developed on the basis of characteristics of closed-domain. Naive Bayes based on multivariate Bernoulli model is improved on the grounds ofthis new catalog. The result of experiments tested on questions gathered during process of instruction shows better promise to this method.
出处
《计算机工程与设计》
CSCD
北大核心
2007年第10期2348-2351,共4页
Computer Engineering and Design
关键词
概念层次网络理论
问句分类
贝叶斯算法
中文信息处理
问答系统
hierarchical network of concepts theory
question catalog
Naive Bayesian
Chinese information processing
questionanswering system
