TDT_CC: A hot topic detection and tracking algorithm based on chain of causes

Abstract

With the development and application of Web3.0, it has become a common social phenomenon that users discuss hot topics on social networks, making them to aggregate into user groups based on the topics, rapidly. The hot topic detection and tracking is helpful for social public opinion supervision and guidance, in addition, it contribute to the user’s behavior mining and analysis. However, users’ interest in some topics often changes as new event occurs, causing the center of hot topics to change over time. For tracking the heat of topic in real-time, we proposed an effective algorithm to detect and track hot topic based on chain of causes (TDT_CC). Firstly, we treat the events as attributes of topic and add them to the structure of the social networks. Secondly, the subgraphs that induced by specific attributes are mined based on the correlation of event-heat-changing attributes and attribute-extended social network structure.