Detection of Online Social Cruelty Attack from Forums

The web provides volumes of text-based data which are stored in online chatting websites like Twitter, Facebook, Blog and Forum etc. Cyberbullying is a social aggressive, powerful negative effects for individuals, specifically adolescents and youngsters. Nowadays, methods for automatic thoughts of mining in the online data are becoming increasingly important. This framework is proposed to extract Cyberbully polarity from the Forum using Fuzzy logic technique. At first, the given input is pre-processed. Subsequently, the pre-processed data will be sent to the features extraction method. Probability of the words are calculated by using Fuzzy Decision Tree Method. Fuzzy rules can be applied in all these features to extract the certain set of cyberbully words like bad words, insulting words, threatening words and terrorism words from the given input. Finally this method will return the reduced and accurate cyberbully words. This method is performed by human annotation, the existing methods like Mamdani Fuzzy System and Naïve Bayes classifier. Extensive experiments are performed by using fuzzy logic on crime debate forum and the results show that this proposed approach is better than the traditional one.