IDENTIFYING POSSIBLE RUMOR SPREADERS ON TWITTER USING THE SVM AND FEATURE LEVEL EXTRACTION

In everyday life, many events occur and give rise to various kinds of information, which are also rumors. Rumors can cause fear influence public opinion about the event in question. Identifying possible rumor spreaders is extremely helpful preventing spread Feature extraction be done expand feature set, consists conversational features form social networks formed from user replies, such as following, tweet count, verified, etc., with text analysis punctuation sentiment values. These become instances used for classification. This study aims identify rumors on Twitter SVM classification model. instance-based algorithm good linear non-linear classification, additional kernels used, linear, RBF, sigmoid. The research focuses getting best model high performance values all models kernel functions that have been defined. It was found RBF has a overall value each data combination ratio amount 1:1 or difference very large. gives accurate results an average 97.02%. With wide distribution data, able map properly.