摘要
近年来,文本情感倾向性分析已成为自然语言处理领域的热点,在垃圾过滤、文本分类、网络舆情分析等领域有广泛的应用。将研究中文文本词句的情感分析问题,重点解决喜、怒、哀、惧四类粒度大的情感分析问题。首先构建喜、怒、哀、惧基准情感词,然后对情感词特征进行分析,进而挖掘潜在情感词,最后使用支持向量机分类的方法融合词特征、词性特征、语义特征等各种特征,对句子进行情感识别及分类。实验表明,在COAE2009评测任务情感词句识别此方法是合理和有效的。
The analysis on text emotional inclination has received much attention from natural language processing field in recent years,which can be widely used in spam filtering,text classification,network public opinion analysis and other applications.This paper presents a method for analysing the emotions on words and sentences in Chinese texts,which focuses on solving four kinds of emotion analysis with big granule including happy,angry,sad and fear.The seed emotional words including happy,angry,sad and fear are firstly set up,and then we analyse the characteristics of emotional words and mine potential emotional words,finally we employ support vector machine to combine the lexical,part of speech and semantic features to recognise and classify the emotions of sentences.Experiment result shows that the method is reasonable and effective when applied to emotional words and sentences recognition in evaluation task of COAE2009.
出处
《计算机应用与软件》
CSCD
2011年第9期225-228,共4页
Computer Applications and Software
关键词
情感词
情感分析
支持向量机
特征选择
Emotional words Emotion analysis Support vector machine Feature selection
