On exploring ambivalent expression in Weibo

Abstract

In the epoch of the Internet, tremendous developments of online social media provide abundant user data, which could vividly reflect users' activities, thoughts, and emotions. As a result, the online data from social media become a popular resource for the research of human being. Among these studies, the emotion research is one of the most important factors for us to understand and predict users' behaviors. However, the sentiment analysis for Chinese short text from social media is non-trivial, due to the instinctive characteristics of Chinese short text. Along this line, this paper focuses on the emotion expression in the Chinese twitter-like social media, Weibo, and proposes an innovative and reliable method to identify the positive, negative and even ambivalent emotion in Chinese tweets. The results on the emotion expression of Chinese Weibo reveal that ambivalent expression is more common in Weibo than in twitter. The sharp contrast is cause the Chinese Culture, which advocates the golden mean and encourages dialectic analysis. As Chinese Weibo users show stronger inclination of passive mood, there is more demand to release their negative emotion. The emotion shift detected in ambivalent tweets and especially the shift from negative to positive mood indicates that the ambivalent tweet is a kind of cognitive reappraisal strategy. Chinese users build new cognitions and different opinions to balance their positive and negative emotion, and finally reach the emotion middle course through the ambivalent tweets. The topic preference analysis also suggests that users who try to regulate their negative emotion and keep balanced mood are more interested in delightful topics like sports or entertainment, and avoid those sensitive topics like economy and politics. All these results help us understand the emotion state and behavior of Weibo users, especially from the perspective of dialectic moderate view.