摘要
在文本的情感倾向性研究中缺乏对多种情感共现的转折句式的有效分析,为此提出一种专门对转折句式进行有效情感倾向性分析的方法。充分分析汉语中转折句式的结构特点,通过已有资源构建中文情感词典、转折词表、否定词表,依据转折句式中转折词、否定词、情感词的组合规律提出用于进行情感分析的启发式规则。在公开语料库的实验中,该方法能更好地对转折句式进行情感倾向性分析,将此规则融入到传统的朴素贝叶斯情感分类模型后,能获得更高的情感分析精度。
Because the lack of effective analysis of transition sentences including various emotions co-occurrences in the research of text sentiment polarity analysis,a method studying the sentiment polarity of transition sentences was proposed.The structure characteristics of transition sentences were analyzed sufficiently,and a Chinese sentiment dictionary,a turning words dictionary and a negative words dictionary were was constructed according to the existing resources,and the heuristic rules used for sentiment analysis were offered according to the combination laws of adversative conjunctions,negative words,and sentiment words in the transition sentences.In the experiments using data of open corpus,the new method can analyze the sentiment polarity of transition sentences better,and after the heuristic rules are added to the traditional nave Bayes emotion classifier,the emotion classifier has higher classification accuracy.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第12期4289-4295,共7页
Computer Engineering and Design
基金
武汉大学软件工程国家重点实验室开放课题基金项目(SKLSE2012-09-30)
山西省自然科学基金项目(2013011015-2)
关键词
情感倾向性
转折句式
情感词
转折词
否定词
sentiment polarity
transition sentence
sentiment word
adversative conjunction
negative word
