基于类熵距离测量的热点话题识别方法研究  

Research on Identifying Method for the Hot Topics Based on Class Entropy Distance Measurement

作　　者：何建民[1,2] 张义[1] 

机构地区：[1]合肥工业大学管理学院,安徽合肥230009 [2]过程优化与智能决策教育部重点实验室,安徽合肥230009

出　　处：《情报科学》2012年第8期1147-1150,1166,共5页Information Science

基　　金：国家自然科学基金资助项目(70631003;70672097);教育部人文社会科学研究资助项目(10YJA630055)

摘　　要：网络社区热点话题识别是监测与引导网络舆情的关键问题,本文针对该问题,通过分析热点话题的属性特征和应用离差最大化、类熵距离法,计算出各属性的权重和话题的正熵、负熵及优劣度,并据此对话题进行排序,找出热点话题。最后经实证验证该方法是合理有效的,所得结果可作为政府监控网络舆情的依据。The online community hot topics recognition is the key problems of monitoring and guiding public opinion, for resolving the issue, in this paper we analyze the hot topic attributes and apply methods of deviation maximization and class entropy distance, calculate the weight of each attribute and the positive, negative entropy of each topic is entropy, and find hot topics according to sorting the topic. Finally the empirical test that this method is reasonable and effective; the results can be the basis for government monitoring public opinion.

关 键 词：网络话题 类熵距离测量 优劣度评价 热点话题识别 

分 类 号：G350[文化科学—情报学] TP393[自动化与计算机技术—计算机应用技术]