基于扩充词典和规则集的突发事件评论情感分类  

Sentiment classification of breaking news reviews based on expanded lexicon and rule sets

作　　者：仲兆满 熊玉龙 黄贤波 ZHONG Zhao-man;XIONG Yu-long;HUANG Xian-bo(School of Computer Engineering,Jiangsu Ocean University,Lianyungang 222005,China;Project Development Department,Jiangsu Institute of Marine Resources Development,Lianyungang 222005,China)

机构地区：[1]江苏海洋大学计算机工程学院,江苏连云港222005 [2]江苏省海洋资源开发研究院项目开发部,江苏连云港222005

出　　处：《计算机工程与设计》2024年第9期2812-2820,共9页Computer Engineering and Design

基　　金：国家自然科学基金项目(72174079);江苏省“青蓝工程”优秀教学团队基金项目(2022-29)。

摘　　要：为准确分析突发事件发生后网民评论文本的情感倾向,提出一种结合扩充情感词典和规则集的情感分析模型。根据词语信息熵筛选领域情感词汇,利用卡方检验判断领域情感词汇的情感极性,得到突发事件领域情感词典;根据文本情感规则集与扩充情感词典计算文本情感值,对低于情感值阈值的文本使用集成学习模型进行二次分类,得到突发事件评论文本的情感类别。通过实验验证了该模型的有效性,为突发事件情感分析提供了可参考的模型和求解算法。To accurately analyze the sentiment tendency of netizens’comment texts after the occurrence of unexpected events,a sentiment analysis model combining an expanded sentiment lexicon and a rule set was proposed.The domain sentiment words were filtered according to the word information entropy,and the sentiment polarity of the domain sentiment words was judged using the chi-square test to obtain the domain sentiment lexicon of breaking events.The sentiment analysis rule set was combined with the expanded sentiment lexicon to calculate the sentiment value of the text,and the text below the sentiment value threshold was classified twice using the integrated learning model to obtain the sentiment category of the critical incident comment text.The effectiveness of the model is verified through experiments,which provides a referenceable model and solution algorithm for sentiment analysis of breaking events.

关 键 词：情感分析 突发事件 情感词典 规则集 信息熵 集成学习 卡方检验 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]