Detecting Hot Topics in Sina Weibo Based on Opinion Leaders

Abstract

Sina Weibo, as one of the most popular and fast growing social network, has gradually become the field where hot topics appear, propagate, and outbreak. In order to discriminate and find out hot topics in micro-blog information, we conduct a series of studies on Sina Weibo, and one of our key findings is that opinion leaders play a very important role in the propagation of hot topics. A smart discriminant model is proposed in this paper to detect hot topics in time, which takes the structure information and propagation characteristics of Sina Weibo as well as the users' influence into consideration. Moreover, word co-occurrence graph is used to extract and display topics. This model has some excellent characters such as a low coupling degree between modules and a low requirement for the amount of data. By experimental verification, it can detect hot topics effectively. Index Terms - Sina Weibo, hot topic, discriminant model, opinion leader, word co-occurrence graph