Detection of Hot Topics Using Multi-view Text Clustering.

Clustering tweets aims to obtain topically coherent grouping of documents i.e. clusters of tweets that can be exploited for multiple applications such as topic detection, news extraction, etc. However, handling such data is considered challenging due to its noisy aspect, the lack of context and the length constraint on one hand, and the natural aspect of text regarding its different interpretations and representations on the other hand. In fact, a single representation model cannot capture the various aspects of text which leads to losing valuable information. Targeting these issues, we propose a multi-view tweets clustering method that exploits various representation models in order to improve the clustering results. The experimental results show that the proposed method enhances the clustering quality.