信息熵方法及在中文问题分类中的应用  

Method of information entropy and its application in Chinese question classification

作　　者：张巍[1,2] 陈俊杰[2] 

机构地区：[1]山西职工医学院信息中心,太原030012 [2]太原理工大学计算机科学与技术学院,太原030024

出　　处：《计算机工程与应用》2013年第10期129-131,179,共4页Computer Engineering and Applications

基　　金：国家自然科学基金(No.60970059);山西省科技攻关项目(No.20110313019);山西省卫生厅科技攻关计划项目(No.2011073)

摘　　要：针对中文问题分类方法中布尔模型提取特征信息损失较大的问题,提出了一种新的特征权重计算方法。在提取问题特征时,通过把信息熵算法和医院本体概念模型结合在一起,进行问题的特征模型计算,在此基础上使用支持向量机方法进行中文问题分类。在城域医院问答系统的中文问题集上进行实验,证明了该方法的有效性,大类准确率及小类准确率分别达到89.0%和87.1%,取得了较好的效果。Aimed at the problem of greater information loss to use Boolean model to extract the feature during Chinese question classification, a new method which calculated feature weight is proposed. When the question feature is extracted, the model of question feature weight is calculated by a combination of information entropy algorithm and hospital ontology concept model. On that basis, the method of Support Vector Machine is used to classify Chinese questions. The classification method is tested on Chinese question set of the city-domain hospital question answering system. This method is proved to be effective and a better result is achieved. Results show that the accuracy of coarse class and fine class achieves 89.0% and 87.1%.

关 键 词：信息熵 本体 问题分类 支持向量机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]