Emotion-based Features In Gender Recognition Of Chinese Micro-blog

With the vigorous development of social media, micro-blog is becoming a hot network communication platform for more and more users. Gender recognition of micro-blog users has become a hot research topic. A lot of efforts had been made on gender emotional differences in the field of psychology and linguistics. Nonetheless, research on gender recognition of Chinese micro-blog users using gender emotional differences is still new. This paper took Chinese micro-blogs as the research object, analyzing the differences of males and females in the perspective of emotion. Based on the differences in the emotion expressions of males and females, a method of gender recognition of Chinese micro-blog users using emotion features was proposed.Firstly, research on emotion recognition of Chinese micro-blog was made, mainly including two aspects. On the one hand, this paper made a comprehensive analysis of emoticons in a large corpus of Chinese micro-blogs, including their usage, distribution and characteristics in emotion expression. After that, this paper established a dimensional vector for a target emoticon, and applied the emotion vectors of emoticons to emotion recognition. The experimental results on two data sets showed that the emoticon vectors effectively improved the precision of micro-blog emotion recognition. On the other hand, this paper arranged Ekman's six emotion classes into a tree structure according to the relationship between emotional polarity and mutual relations. Based on this tree structure, an approach of hierarchical emotion analysis of Chinese micro-blog was proposed. The experimental results showed that hierarchical classification method reduced the effect of the highly imbalanced distribution of the emotional categories on the classification results, and improved the accuracy of micro-blog emotion recognition.Secondly, this paper analysed the differences of males and females in the perspective of emotion. Based on the differences in the emotion expressions of males and females, a method of gender recognition of Chinese micro-blog users using emotion features was proposed. The emotion features included Ekman's emotion features, emotional words and linguistic style features associated with emotions. The experimental results showed that using emotion features improved the accuracy of gender recognition.Finally, this paper explored how to establish gender bias dictionary. First, extract gender bias candidate words from micro-blog texts. Then, based on the significant differences in the frequency of the candidate words, choose gender bias words from the candidate words. If a candidate word was used more frequently by male users than female users, the word would be a male bias word. Otherwise, it would be a female bias word. Last, use the established gender bias dictionary to recognize the gender of Chinese micro-blog users.