Design of Opinion Leader Identification System for Hot Public Opinion

Opinion leaders are key figures in the development of public opinion. Their published content and emotions often affect other users and dominate the development trend of the entire hot public opinion. The opinion leader identification system designed in this paper can automatically discover potential opinion leaders and predict hot public opinion. The system automatically acquires real-time hotspot public opinion data, uses the K-means++ algorithm to cluster users, finds existing opinion leaders, and then transfers the knowledge of opinion leader nodes to ordinary user nodes by constructing KTGNN, enriching the node representation of ordinary users, mining Identify potential opinion leaders among them. In the actual public opinion data test, the system can effectively discover potential opinion leaders and help predict hot public opinion trends.