AN ENHANCED DEEP LEARNING FRAMEWORK FOR FAKE USER IDENTIFICATION IN SOCIAL NETWORKING SITES

Abstract

Online Social Networking Sites (OSNs) such as Facebook, Twitter (X), Instagram, and LinkedIn have become an integral part of modern communication, enabling users to share information and build social connections. However, the rapid growth of these platforms has also increased the number of fake user accounts that spread spam, phishing attacks, misinformation, and malicious content. Detecting fake users is challenging due to evolving attack strategies, diverse user behavior, and the large volume of social media data. This paper proposes an Enhanced Deep Learning Framework for Fake User Identification by integrating content-based analysis, behavior-based analysis, and machine learning techniques. The framework utilizes TFIDF for text feature extraction, Vector Space Model (VSM) with Cosine Similarity for content similarity analysis, and Random Forest for user classification. Bloom Filters are employed for efficient data retrieval and duplicate detection, while behavioral analysis examines user posting patterns, account activity, and network interactions to identify suspicious accounts. By combining textual and behavioral features, the proposed framework improves fake user detection accuracy while reducing false positive rates. Experimental results demonstrate enhanced classification performance, faster processing, improved scalability, and greater robustness compared with traditional detection methods.