摘要
移动互联网的快速发展使得网络上数据量剧增.如何从纷繁复杂的信息中提取出对人们有用的信息就成为一个亟待解决的课题.本文提出一种改进的基于情感词典的倾向分析方法,该方法在情感词典中加入领域情感词,并且通过构建辅助词典集来进行辅助分析.同时给出了一种半自动的词典维护方法来发现新词和更新词典集.通过对手机领域的评论进行文本级的情感倾向分析,正面情感分析的准确率和召回率达到0.713和0.769,负面情感的准确率和召回率达到0.738和0.706,与传统基于情感词典的方法相比准确率和召回率都有较大提高.
The volume of data on the network increases remarkably with the rapid development of mobile network. How to extract useful information from the complicated information becomes an urgent problem to be solved. An improved semantic orientation analysis was proposed, in which the sentiment dictionary was expanded with domain emotional words, and an auxiliary dictionary set was constructed for assistant analysis.To maintain all the dictionaries, a semi-automatic method was also presented. The results in the domain of mobile phone show that the rates of accuracy and recall of positive emotion analysis are 0.713 and 0.769, and the rates of accuracy and those of negative emotion analysis are 0.738 and 0.706, which are improved compared with using the traditional method.
出处
《武汉工程大学学报》
CAS
2015年第10期45-50,共6页
Journal of Wuhan Institute of Technology
基金
国家自然科学基金(61272115
60873024)
关键词
情感倾向分析
情感词典
辅助词典集
semantic orientation analysis
sentiment dictionary
assistant dictionary set
