一种消除情绪孤立点的中文微博情绪分析

Analysis of Chinese Micro-blog Emotion Based on Emotion Outliers Elimination

随着互联网的迅速发展,微博已经成了人们抒发个人情绪的重要平台,对微博文本的意见挖掘和情绪分析也受到了大量学者的关注和研究。其中,由于微博主题的发散性以及情绪的多元性,导致微博文本中出现了大量的情绪孤立点,干扰了微博情绪的判断。因此,论文提出通过情绪相似度的方法来消除语料中的情绪孤立点,并利用规则方法来判断微博文本的情绪。实验表明,消除语料中的情绪孤立点有效地提高微博情绪分析的准确性和精确率。

With development of the Internet, micro-blogs has become an important platform for people to express per- sonal emotions and feelings. Therefore, micro-blog attracts more and more experts and scholars for opinion mining and senti- ment analysis. But the wide range of topics and the diversity of expressions of emotion in micro-blogs lead to quite a few isolated point（emotional outliers）. In the same time, these emotional outliers interfere the emotional judgment in miero-blogs. Therefore, this paper calculates the emotional similarity of emotional words based on corpus context, and uses it to remove e- motional outliers. Finally, based on rule approach, the emotions of Chinese micro-blogs are analyzed. The experimental re- sults show this approach is effective in identifying the major emotion of micro-blogs, and improves the accuracy of micro-blog emotional analysis.