AUTOMATIC IDENTIFICATION OF NEWS TWEETS ON TWITTER

Social media is most popular medium for its real-time and rapid information dissemination. Twitter dominates the overall social community and supports sharing of news and one's personal opinion in the form of news and personal tweets respectively. Personal tweets share one's opinion about a happening while news tweets communicate the facts of the happenings. Hence the summary of news tweets excluding personal tweets provides unbiased current fact situation of the dynamically changing world. The authors propose an effective methodology for identifying news tweets by considering the sources that generate the tweet in addition to content of the tweet. In first phase, the methodology collects tweets generated from popular sources using screenNames that partially match with an identified set of patterns. The second phase independently identifies candidate news tweets based on the occurrences of news words in their content. Finally, the tweets obtained from the previous two phases are merged and screened to obtain an accurate set of news tweets. The improved accuracy of the proposed methodology in identifying news tweets on a wide range of topics substantiates the impact of tweet sources for collecting news tweets.