Abstract

Currently, online social media platforms serve as the most prevalent and rapid means of exchanging information in the realm of technology. Individuals from diverse backgrounds predominantly devote a significant amount of their time to engaging with social networking sites. A vast quantity of information is generated and disseminated globally via social networks. These incentives have led to unauthorized individuals carrying out hostile actions against users of the social site. The creation of false accounts on social media is seen as more detrimental than any other kind of cybercrime. The detection of this infraction must occur prior to informing the customer about the creation of the fraudulent identity. Several algorithms and methodologies have been proposed for detecting fraudulent identities, mostly leveraging the huge volume of unprocessed data generated by social networks. This study aims to identify fraudulent identities on social media accounts using a dataset from Twitter. Diverse machine learning methods have been used to assess the suggested outcomes using natural language processing approaches. SVM, Fuzzy Random Forest, and Naïve Bayes have been used for classification purposes. The empirical study demonstrates the efficacy of the system and its superior accuracy compared to other machine learning algorithms and preexisting systems.Keywords: Machine Learning, Naïve Bayes, Fuzzy Logic, Random Forest, twitter dataset, fake identity, bots, Natural Language processing, classification.