Detecting opinion evaluation object drift in Chinese microblog posts

Abstract

ABSTRACTThe drift of opinion evaluation object can reflect the evolution of online public opinion from a fine‐grained level. By representing and calculating the distribution of opinion evaluation objects, we propose a method to detect its drift. A dataset of posts published by opinion leaders in social‐events discussion is collected from a microblog application in China. The preliminary result on the dataset shows that when subject discussion changes, opinion leaders will focus on the identity characteristic of one person in most cases. By using our proposed method, the drift of opinion evaluation object can be clearly uncovered. Furthermore, by combining with opinion sentiment tendency drift, our research can enrich the perspective of public opinion evolution analysis and better the monitoring of public opinion crisis.