Detection of Anomalies in Social Multimedia Domain With AI Integrated Software‐Defined Networking

Social multimedia (SM) platforms such as Instagram, LinkedIn, YouTube, Facebook, and Twitter enable users to communicate, share ideas, and exchange information within virtual networks. The widespread use of multimedia‐based applications has led to a substantial increase in social media traffic, often containing sensitive personal and interaction data. These data are vulnerable to security threats, including identity theft and information misuse, highlighting the need for effective anomaly detection mechanisms. This research proposes an enhanced binary spotted hyena optimized dependent linear regression (EBSHO‐DLR) method to detect anomalies in SM networks, ensuring secure data transmission. The framework begins with collecting a Twitter dataset from India, followed by pre‐processing using Gaussian blur (GB) to filter anomalies and histogram of oriented gradients (HOG) for feature extraction, removing irrelevant information. The proposed method integrates software‐defined networking (SDN) to provide dynamic, programmatically efficient network management, improving performance and monitoring compared to conventional approaches. Experimental evaluation demonstrates the effectiveness of the method in detecting fraudulent activities such as identity theft, profile cloning, and unauthorized data collection. Key performance metrics include precision (96.7%) and accuracy (97.8%), confirming the robustness and reliability of the EBSHO‐DLR approach for securing social multimedia communication systems.