Tweets clustering using latent semantic analysis
Norsuhaili Mahamed Rasidi, Sakhinah Abu Bakar, Fatimah Abdul Razak · AIP conference proceedings · 2017
Social media are becoming overloaded with information due to the increasing number of information feeds. Unlike other social media, Twitter users are allowed to broadcast a short message called as ‘tweet”. In this study, we extract tweets related to MH370 for certain of time. In this paper, we present overview of our approach for tweets clustering to analyze the users’ responses toward tragedy of MH370. The tweets were clustered based on the frequency of terms obtained from the classification process. The method we used for the text classification is Latent Semantic Analysis. As a result, there are two types of tweets that response to MH370 tragedy which is emotional and non-emotional. We show some of our initial results to demonstrate the effectiveness of our approach.