Fake Profile Detection on Social Media Using Hybrid 2D CNN and AES-BiLSTM with Network Analysis

Abstract

Social media platforms transform communication by enabling users worldwide to connect, share and interact effortlessly. With thousands of new users joining daily these platforms generate an immense volume of data including text, images and videos. While this growth fosters engagement and connectivity, it also presents challenges such as the proliferation of fake accounts and online impersonation. Malicious profiles often exploit these networks to distribute malware, viruses and harmful URLs, undermining the trust and security of social media ecosystems. Detecting and eliminating fake profiles is crucial to maintaining the integrity of these platforms and ensuring user safety.This paper proposed a hybrid artificial intelligence model to address these challenges by combining advanced preprocessing, feature extraction and classification techniques. During the preprocessing phase, the model removes duplicate entries, imputes missing values using mean imputation and scales the data with the Min-Max normalization technique to ensure consistency. For feature extraction, Principal Component Analysis (PCA) is used to decrease the dimensionality of the data, ensuring efficient processing. The hybrid classification framework integrates a 2D Convolutional Neural Network (2D CNN) to extract spatial features from the input data. Additionally, an Attention-Enhanced Stacked BiLSTM (AES-BiLSTM) is utilized to capture temporal features, while the added attention mechanisms improve the model's focus on the most relevant information.The 2D CNN and AES-BiLSTM models are further enhanced through hybrid optimization, with their hyperparameters fine-tuned using the Seagull Optimization Algorithm (SOA). Implemented in Python, this approach achieves a high accuracy of 98.9%, precision of 98.9%, specificity of 98.9%, and an F1-score of 99%. It provides a robust and scalable solution for detecting fake profiles on social media platforms. By addressing the limitations of traditional methods, this model offers an effective framework to safeguard social media ecosystems against fraudulent activities.