Research on Predicting Emergent Online Public Opinion Topics Based on Dynamic Opinion Leaders

Exploring the changes in topics discussed by netizens during sudden incidents of online public opinion, predicting the development trend of event topics, can promote targeted public opinion guidance and achieve the goal of comprehensive control and precise guidance. This is of great significance for the management of online public opinion in the new era. This study selected sudden public opinion events on the Weibo platform, constructed a social temporal network, calculated influence using user interaction data, and introduced memory effect parameters based on the Ebbinghaus forgetting curve to construct an IFd algorithm for dynamically identifying opinion leaders based on memory effect. Using LDA topic model to construct user topic maps, calculating the number of opinion leaders under each topic, identifying popular and niche topics, drawing topic word frequency heat maps, predicting potential hot topics, and verifying prediction accuracy based on real data. The research results indicate that this method can effectively predict the direction of online public opinion topics, providing a method and means for dynamic supervision of online public opinion. It is of great significance for guiding public opinion direction and taking corresponding measures in a timely manner.