Detecting the Anti-Social Activity on Twitter using EGBDT with BCM

Abstract

The rise of social media and its consequences is a hot topic on research platforms. Twitter has drawn the attention of the research community in recent years due to various qualities it possesses. They include Twitter's open nature, which, unlike other platforms, allows visitors to see posts posted by Twitter users without having to register. In twitter the sentiment analysis of tweets are used for detecting the anti-social activity event which is one of the challenging tasks in existing works. There are many classification algorithms are used to detect the anti-social activities but they obtains less accuracy. The EGBDT (Enhanced Gradient-Boosted Decision Tree) is used to optimize the best features from the NSD dataset and it is given as input to BCM (Bayesian Certainty Method) for detecting the anti-social activities. In this work, tweets from NSD dataset are used for analyzing the sentiment polarity i.e. positive or negative. The efficiency of the proposed work is compared with SVM, KNN and C4.5. From this analysis the proposed EGBDT and BCM obtained better results than other techniques.