Prediction of Emotion Trend of Weibo Users Based on Multi-Level Attention Mechanism

Abstract

With the advent of the open information age, Weibo, as a carrier of data sharing and information services, has become a platform for user information and opinion release and emotional feedback under the popular service mode. When some users express their emotions through Weibo, it is more peaceful. When they need to express their intense emotions, the proportion of user participation will be reduced. Although the communication on the Internet is often regarded as virtual, this kind of communication which is not limited by time and space will also produce different degrees of interaction between people. In depth understanding of the generation mechanism of users' emotional behavior can deepen the understanding of the formation of group emotional behavior in Weibo network, which has a certain reference value for Weibo's public opinion control and Weibo marketing. Deep learning theory is introduced to analyze the emotion of Weibo user group based on multi-level attention mechanism to better grasp the emotional information in the text and improve the success rate of emotion trend prediction.