A Hybrid Approach for Enhanced Detection of Fake Profiles in Online Social Networks

The proliferation of information sharing on social media platforms has attracted attackers who often use fake profiles to commit cybercrimes. This work presents a hybrid methodology that addresses the limitations of current techniques by integrating the advantages of deep learning and machine learning algorithms. The suggested technique employs an improved feature extraction layer and a hybrid classifier to optimize the performance of false profile detection. The methodology synergistically integrates the advantages of content-based and behavioral-based analysis methodologies, utilizing a wide range of features that improve accuracy. Content-based analysis involves the examination of textual and multimedia content linked to user profiles. It utilizes natural language processing to identify abnormalities and conflicts. The behavioral-based study centers on examining user interactions, utilizing machine learning algorithms to identify trends in user activity, relationships, and posting behavior. Experimental studies carried out on real data sets obtained from Twitter demonstrated a notable enhancement in the precision of detection when compared to conventional approaches. A combination of these methods shows potential for improving the security and reliability of online social networks, offering an essential tool for platforms and users to reduce the dangers linked to fake profiles.