Detecting Hot Topics in Chinese Microblog Streams Based on Frequent Patterns Mining

摘要

Microblog plays a more and more important role on the emerging and propagation of the public opinion on the Web. Although topic detection has long been a hot research topic, the characteristics of microblog make it a non-trivial task. In this paper, we propose a novel hot topic detection approach based on keyword extraction and frequent patterns mining. We analyze the characteristics of hot topic microblogs and the topical keywords are extracted according to the increasing rate and frequency in Chinese microblog streams. Different from traditional clustering based detection methods, in this paper we treat the short texts of microblogs as transaction items, and apply Apriori algorithm to generate the hot topics. The experiments in the real dataset verify the efficiency and effectiveness of our proposed methods.