Micro-blog hot event detection based on dynamic event model

Abstract

Micro-blog is a rapid developing platform, where users can share information about the event happened in the real life, and sometimes many hot events even originate in this platform. So it is very important to detect the hot event in the micro-blog texts. Micro-blog text is a kind of the short, fractional and grass-root text. Based on the analysis of the dynamic evolvement property of the event, we propose a micro-blog event detection method based on dynamic event model, where the event model is created based on the two centroids. This paper firstly analyzes the dynamic evolvement property, secondly introduces the creation and modification process of the two-centriods based event model, and finally gives the event detection method based on the proposed dynamic event model. Experimental results showed that our proposed model improves the performances of the micro-blog event detection greatly. © 2013 Springer-Verlag Berlin Heidelberg.