Topic Based Temporal Generative Short Text Clustering

Abstract

In Social network paradigm, Twitter is the most widely used microblog nowadays. In the microblog environment, content or the specific theme for tweets are identified using hashtags but in general, all tweets do not contain hashtags. As the tweets form and spread very fast in microblogs, recommending relevant hashtags to tweets is emerging as a challenging task. This paper proposed an approach for short text clustering using temporal classification approaches. Hence for the batch of tweets both LDA and HDP topic modeling are attempted. In this paper, a hashtag is recommended for each tweet for mapping the topics obtained and the topic with the higher probability is considered as the hashtag of that tweet.