A Comparative Study of Different Sentiment Analysis Classifiers for Cybercrime Detection on Social Media Platforms

Abstract

In the current scenario, social media has made it very easy to access and exploit the different types of data from various social media platforms, which are freely available to everyone to share their opinions openly. With this open access, the privacy and security of all social media users is a cause of concern and matters. Sentiment analysis plays an essential role in social media security as it is used by many applications domains like risk management, anomaly detection and disaster relief. The article proposes the sentiment analysis approaches and methods for social media defence and assessment. A study on the security challenges related to the user security breach, e-commerce websites, fake news, cyber-bullying, credibility on social media is presented in this paper. Here, the major challenge under consideration is phoney news detection on several social media platform. A brief outline of two classifiers, namely Multinomial Naïve Bayes and Passive-Aggressive Classifier, is presented initially. Then a comparison of two classifiers is conducted to analyze the techniques and methodologies on the Twitter dataset. The results from both the classifiers are discussed based on the performance metrics and the accuracy score. Finally, the article derives some important conclusion and future direction in this domain.