A Framework for Detecting and Tracking Religious Abuse in Social Media

Religious abuse in social media is a common phenomenon in recent years. This abuse affects our religious morality which generates wrong ideas to the people connected in social media about different religious communities. There are many approaches in the state-of the-art have been used to filter out the desired sentiments from the social media. This paper presents such a framework for identifying and tracking suspicious users who mislead people by spreading conflicting religious information. The framework is designed and trained with diverse religious keywords so as to process real time data stream. We used Twitter social data stream for experiment purposes. Initially, the interested tweets (i.e., Islamic) have been extracted; then the preliminary stages of misinterpreted religious data is detected on the basis of Twitters spam policy, user-based and content-based features. Moreover, a web crawler is constructed with API method provided by the Twitter. Afterward, the processed data is classified to distinguish the mistrustful behaviors from regular events (i.e., the trained classifier is applied to the entire data set). Finally, the performance of the framework is evaluated with different classifiers i.e., Support Vector Machine (SVM), Random Forest and Decision Tree classification algorithms. The result shows much closer accuracy of each in most of the cases. However, we can say that the proposed approach can detect and trace any religious fraud acts in social network.