Online interactions are seriously threatened by the widespread fraudulent activity on social networking sites, which is typified using bogus identities and misleading viewpoints. The objective of the present work is to enhance Security Protocols and Digital Literacy on Social Media by proposing a robust anomaly detection model for social media. The present study proposes a comprehensive strategy that integrates technical, legal, and user-focused remedies such as content filtering, algorithmic transparency, and identity verification. When examining anomalies in social media data, Large Number of Rare Events models specifically, the left-skewed log-normal distribution become an invaluable statistical tool. The study uses Large Number of Rare Events models with equations related to vocabulary size and frequency class to solve the problems caused by uncommon events, such as cases of erroneous identities. Validation and application of LNRE models in social media data by utilizing text-based mathematical analysis to tackle fraud detection to be performed in the proposed work. The analysis of lexical diversity metrics offers language richness and a deeper understanding of the content under examination, aiding in the identification of focus areas, opinions, and discussions within the analysed text.