Social Sentiment Detection of Event via Microblog

Sentimental analyses of the public have been attracting increasing attentions from researchers. This paper focuses on the research problem of social sentiment detection, which aims to identify the sentiments of the public evoked by online microblogs. A general social sentiment model is proposed for this task. The general social sentiment model combining society and phycology knowledge are employed to measure social sentiment state. Then, we detail computation of sentiment vector to extract sentiment distribution of blogger on event. Besides, social state for events are computed based on the general social sentiment model and sentiment vectors. Furthermore, we certify that social sentiment are not independent but are correlated with each other heterogeneously in different events. The dependencies between sentiments can provide guidance in decision-making for government or organization. At last experiments on two real-world collections of events microblogs are conducted to prove the performance of our method.