基于分类关联规则的微博情绪分析  

Emotion analysis in Weibo based on class association rules

作　　者：刘思[1] 朱福喜[1] 阳小兰[2] 刘世超[1] 

机构地区：[1]武汉大学计算机学院,湖北武汉430072 [2]武昌理工学院信息工程学院,湖北武汉430023

出　　处：《计算机工程与设计》2016年第12期3361-3365,3393,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61272277);中央高校基本科研业务费专项基金项目(274742);湖北省自然科学基金项目(2014CFB356)

摘　　要：针对微博文本语法不规则、句子间文本联系不紧密的问题,提出一种基于分类关联规则的情绪分析方法。获得一篇微博中相邻句子间的连接词,分别采用KNN和SVM算法对微博中的每个句子进行情绪分析,获得对应的情绪标签;将获得的情绪标签和连接词转换为关联规则项集,通过关联规则挖掘算法获得相应特征;采用SVM算法对获得的特征进行情绪分类,得到整篇微博的情绪类别。实验结果表明,该方法在情绪分类上具有较好的效果。For the irregularity and poor association between sentences of Weibo texts,an approach based on class association rules for emotion classification was proposed.The connection word between adjacent sentences was got,and two emotion labels for each sentence in a Weibo text were obtained using KNN and SVM algorithm respectively.The connection word and emotion label were converted into association rule sets,and features were derived using class association rules mining algorithm.SVM algorithm was adopted to classify the new features to get the whole Weibo emotion category.Experimental results show that this method is effective in emotion classification.

关 键 词：情绪分析 文本分类 K最近邻算法 支持向量机 分类关联规则 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]