基于情感词典与语义规则的微博情感分析  

Microblog Sentiment Analysis Basing on Emotion Dictionary and Semantic Rule

作　　者：陈国兰[1] 

机构地区：[1]南京邮电大学图书馆,江苏南京210003

出　　处：《情报探索》2016年第2期1-6,共6页Information Research

基　　金：江苏省教育厅高校哲社基金项目"众包模式下知识转移的实现机制研究"(项目编号:2014SJB025);南京邮电大学党建与思想政治教育研究校级规划项目"微博时代高校网络舆情的监测与引导--以南京邮电大学为例"(项目编号:XC214020)成果之一

摘　　要：[目的/意义]基于机器学习的中文微博情感分析方法存在处理过程复杂、判断准确率低等问题,探讨新的情感分析方法对该领域研究有实际意义。[方法/过程]首先基于已有的开源情感词典和微博文本的特点,构建一个适合微博文本分析的情感词典;然后基于常见的微博情感句,以微博情感词为中心,对微博文本进行情感语义规则分析,进而提出微博语句文本的情感计算方法;最后基于已标注的微博语料,构建相关词典,包括程度副词词典、否定词词典、关系连词词典。[结果/结论]通过与支持向量机(SVM)的情感分类方法进行实验对比,证明本研究提出的微博情感分析方法的适用性较强。[Purpose/significance]Machine learning-based Chinese microblog sentiment analysis has issues in complex process and low correct rate of judgments, and discussing new emotion analysis method is of practical significance for research in this field.[Method/process]The paper firstly bases on existing open-source emotion dictionary and characteristics of microblog text, constructs an emotion dictionary suitable for microblog text analysis; then it bases on common micro-blog sentiment sentences and takes microblog emotional words as the center, to analyse microblog text’s emotion semantic rule, and puts forward emotion calculation method of microblog statement text; lastly it bases on annotated microblog corpus to construct related dictionaries including of degree adverbs dictionary, negative words dictionary, and relation conjunctions dictionary.[Result/conclusion]Experimental comparison between it with sentiment classification method of SVM shows that microblog emotion analysis proposed in this study is more suitable.

关 键 词：微博 微博文本 情感分析 情感词典 语义规则 句级情感分析 

分 类 号：G250.23[文化科学—图书馆学]