Topic extraction in social media

Social networks have become the most important source of news and people's feedback and opinion about almost every daily topic. With this massive amount of information over the web from different social networks like Twitter, Facebook, Blogs, etc, there has to be an automatic tool that can determine the topics that people are talking about and what are there sentiments about these topics. The goal of the research described in this paper was to develop a prototype that can "feel" the pulse of the Arabic users with regards to a certain hot topic. Our experience in extracting Arabic hot topics from Twitter is presented in this paper. The unigram words that occurred more than 20 times in the whole corpus were used as features for clustering the tweets using bisecting k-mean clustering algorithm. This has resulted in purity of 0.704 and entropy of 0.275. The score generated for the quality of the generated topic was 72.5%.