Web Event Topic Analysis by Topic Feature Clustering and Extended LDA Model

To analyze topics of a large number of web events, we proposed an event topic analysis approach by topic feature clustering and extended LDA (latent dirichlet allocation) model. The extended LDA model is dimension LDA (DLDA) which integrates topic probability of LDA model. We represent an event as a multi-dimensions vector and use DLDA model to select topic feature words in events. We aggregate events which have a common topic by topic feature clustering. In clustering process we use dynamic Kmeans method to automatically select suitable number of clusters. In this paper a topic term generating rule is proposed to compose topic terms by clustered topic feature words. We accurately detect a common topic from lots of different events and analyze topic terms for events. Experiments on dataset results show that the web event topic analysis approach has high accuracy.