Unveiling Malicious Intent: A Theoretical Framework for Detecting Anomalies in Social Media Behaviours

Abstract

The proliferation of social media platforms in business and politics has, unfortunately, facilitated the emergence of undesirable behaviours. From personal vendettas to corporate and political propaganda, engagement on social media has incentivised actions that are detrimental to individuals and society at large. The anonymity of hacked profiles has encouraged individuals to spread criticism and misinformation with impunity. This study delves into social media data to uncover the intricate relationships that underpin these problematic behaviours. Employing a sophisticated system, it identifies aberrant behaviours by detecting malicious or fraudulent accounts within social networks. Drawing on theories such as Influence, Homophily, and Balance Theory, a theoretical social framework is employed to enhance the accuracy of classifying potentially harmful users. Utilising metrics like the Jaccard coefficient, the system evaluates the similarity between user behaviours, incorporating graphical and linguistic cues to categorise end-users. The framework’s efficacy is rigorously assessed using standard parameters, including confusion matrices, to gauge performance. Additionally, the study recommends employing a friend connection identification framework for enhanced social atom anomaly detection, further fortifying efforts to combat undesirable behaviour on social media platforms.