WebShield 5.0: Harnessing AI and NLP to combat web threats in Industry 5.0

Industry 5.0 characterized by the integration of human intelligence and advanced technologies is inherently more connected and interdependent than previous industrial paradigms. This increased connectivity exposes to various web-based attacks and calls for strong security controls. To address the challenges and enhance attack detection, this paper introduces Ingress Manager (IM), a novel approach that amalgamates Natural Language Processing (NLP) with Machine Learning (ML) to mitigate web-based threats. By combining multimodal data and utilizing the Mayfly optimization algorithm for feature selection, IM carries out a thorough analysis for efficient web-based attack detection. Mayfly Optimization is considered to be a variation of Particle Swarm Optimization (PSO), combining the benefits of Firefly Algorithm, Genetic Algorithm (GA), and PSO. Experiments on the HTTP CSIC-2010 dataset show that this integration is effective, as evidenced by the notable gains in accuracy, precision, and F-score above baseline models. Notable performance metrics such as accuracy of 98.5753% along with thorough component analysis (ablation study) add deeper understanding to the proposed approach. The paper's contributions lie in its utilization of Industry 5.0 principles, the incorporation of Mayfly optimization for feature selection, and the innovative combination of NLP and ML for robust web-based attack detection.