Microblog-oriented hot topic discovery and trend analysis
Chengfang Tan1,2, Caiyin Wang2, Lin Cui1,2
COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12C) 516-520
1School of Information Engineering, Suzhou University, Suzhou 234000, Anhui, China
2Intelligent Information Processing Lab, Suzhou University, Suzhou 234000, Anhui, China
Microblog has become an important means for people to obtain the first message. This paper proposes a method for hot topic detection and topic trend analysis in the massive microblog short text data set. Firstly, considering the microblog special style and clustering efficiency, we make appropriate improvements on the classical single -pass clustering algorithm to enhance the quality of clustering, and then put forward the topic heat evaluation method and rank topics of each topic cluster based on this method. Finally, in each time window, we calculate the strength of topics depending on the topic strength calculation formula, and discovery the topic evolution trend. Experiments show that the proposed method can not only effectively detect hot topics, but also can clearly find out the topic evolution process.