Sentiment Analysis Of Tibetan Weibo Based On Multi-feature Fusion

With the development of Internet technology,the improvement ofinformation processing technology and the promotion of social networking applications,more and more users tend to express their views on the current affairs and hot events through microblogging.At present,the sentiment analysis is mainly aimed at Chinese and English,while the research of Tibetan sentiment analysis is in infancy.With the increasing of Tibetan microblogging users,Tibetan microblogging sentiment analysis has become particularly urgent and important.Understanding emotion of the Tibetan has great significance to maintain the relationship between Han and Tibetan relations,control the public opinion and predict emergencies.For the Tibetan Microblogging Sentiment Analysis,the paper mainly studies the following:(1)For the lack of Tibetan sentiment dictionary resources,the paper proposes a kind of sentiment dictionary expansion and filtering technology based on large-scale corpus,and constructs a Tibetan sentiment dictionary with a relatively high emotional quality.(2)For the sparse problem of sentiment characteristics of microblogging,the paper designs the multi-feature fusion sentiment classification method which selects features from sequence rules,semantics and emoticons.By adjusting the feature combination strategy,the macro accuracy of the classification experiment finally reached 88.26%.