Keyword-based semantic analysis of microblog for public opinion study in online collective behaviors

Abstract

This study employs text mining and semantic analysis methods to analyze the change of public opinion in the formation of online collective behaviors in crisis management. A case that related to law enforcement violence is investigated to study the communication between government and the public. The development of event is framed based on news data. The paper conducts a semantic analysis on microblog data and extracts high-frequency keywords and co-occurrence words at each stage of the event. By comparing key words at different stages, the paper proposes a new way to gain insight into the requirement of the public and predict the change of public opinion.