-
Views
-
Cite
Cite
Na Su, Shujuan Ji, Jimin Liu, Real-Time Topic Detection with Dynamic Windows, The Computer Journal, Volume 63, Issue 3, March 2020, Pages 469–478, https://doi.org/10.1093/comjnl/bxz042
- Share Icon Share
Abstract
Microblog is a popular social network in which hot topics propagate online rapidly. Real-time topic detection can not only understand public opinion well but also bring high commercial value. We design a method for real-time microblog data analysis in order to detect popular long lasting events as well as emerging events. Firstly, a mining frequent items algorithm on microblog data stream is proposed to count approximate word frequency. This mining frequent items algorithm can find the frequent words for some time. Secondly, the windows size of the monitored words is adjusted dynamically according to the duration time and the evolution of events. Lastly, new topics and trends of existing topics can be detected by using dynamic clustering algorithm based on vector space model. Experimental results show that the proposed algorithms can improve performance in terms of running time and accuracy.