Social networks analysis based on topic modeling

Abstract

Understanding the discussed content in social networks is an uneasy problem but brings a lot of advantages for different fields, such as marketing, education, social trends, security. To build up the system for supporting products marketing in social networks, we develop models of content-based social networks analysis in order to find out the discussed topics. The system consists of steps, such as extracting messages, discovering and automatically labeling the discussed topics, in which we pay attention to time factor. Experimented with the Enron corpus containing 11,945 e-mails discussed by 147 users and estimated 50 topics, the system has found out many useful topics and opened new research and application directions.