Finding information spreading and topics based clustering on Twitter

Abstract

Online social networks are ways of spreading news. Some issues in online social networks make diffusion based on lots of reasons and influence is one of those reasons. To distinguishing influence in online social newtorks, statistical techniques are used. Also, diffusion of topics in online social networks are limited by clusters in those networks. Shuffle test is a statistical method to figure out influences in online social networks. Otherwise, Jaccard index is used to find out the clusters in online social network services. Influeces in online social networks are differenct depending on the situation. Also, it is hard to spread news toward other clusters which is based on topics because people tend to mention topics only which they have interests in. We have crawled the entire Korean Twitter and get 3 billion tweets from 6.6 million users. We find out limitation of influences and clusters in online social networks with 19 keywords from political issues, entertainments issues, the name of electronic devices and other general issues. This is the work which shows the pattern and limitation of influences in Korean Twitter across topics.