情感词汇Ontology驱动的话题倾向性计算  

Topic semantic orientation compute based on sentiment words Ontology

作　　者：王晓东[1] 刘倩[1] 张征[1] 

机构地区：[1]河南师范大学计算机与信息技术学院,河南新乡453007

出　　处：《计算机工程与应用》2011年第27期147-151,共5页Computer Engineering and Applications

基　　金：河南省科技攻关计划项目(No.102300410198;No.082102210007;No.072102210063)

摘　　要：针对"未然态"的舆情信息,挖掘网络热点、焦点及敏感话题,把握舆情动态,提高处置与监管网络突发事件能力等,是舆情分析的重要研究内容。对基于情感词汇Ontology的话题倾向性进行了研究。通过计算与情感词汇Ontology中情感词汇的语义相似度、统计话题语料中情感特征词汇的词频,计算语料中情感特征词汇的倾向性权重;根据情感特征词汇的倾向性权重计算话题倾向性强度和整体倾向性。最后在情感词汇Ontology指导下对话题中每篇语料的情感分类和倾向性强度进行规范化细粒度标注。In allusion to the unlighted public opinion information,the important aspects of public opinion analysis are as follows：network hot spots,focus and sensitive topic mining,public opinion trends control,the ability to handle and monitor the network emergency improvement and so on.Topic semantic orientation compute based on sentiment words ontology is studied.The orientation weights of the words are calculated by calculating the semantic similarity of the words in sentiment words ontology and counting the sentiment words frequencies in the topic corpus.Then the topic orientation extent and the overall orientation can be calculated according to the orientation weights of the words.Finally,the sentiment classification and orientation extent of each corpus in the topic is tagged by the fine-grained standard based on sentiment words ontology.

关 键 词：舆情 话题倾向性 情感词汇Ontology 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]