A Lightweight, Explainable Spam Detection System with Rüppell’s Fox Optimizer for the Social Network X

Effective spam detection system is an essential in online social media networks (OSNs) and cybersecurity, that directly influencing the quality of decision-making pertaining to security. Conventional machine learning (ML) methods, including Decision Trees (DT), K-Nearest Neighbors (KNN) and Logistic Regression (LR), were utilized in tackling this issue. Despite their effectiveness, such methods frequently encounter known as "black box" problem, an interpretability deficiency that constrains its deployment into security applications, which comprehending the rationale of classification processes is crucial for efficient threat evaluation and response strategies. To overcome this limitation, this study employs concepts of Explainable Artificial Intelligence (XAI) to propose a interpretable model for X spam account detection. This work introduces an innovative spam detection system utilizing ensemble learning AdaBoost (Adaptive Boosting) method augmented by a carefully crafted framework. The approach employs clean data with data preprocessing, feature selection using a swarm-based, nature-inspired meta-heuristic Ruppell's Fox Optimizer (RFO) Optimization Algorithm which for the first time applied to Cybersecurity and for interpret model prediction Shapley values are computed and illustrated through swarm and summary chart. The model attained a notably accuracy of 99.35% along with high precision, recall and F1-score, surpassing conventional algorithms including DT, KNN and LR in all performance metrics. The results validate the efficacy of the suggested approach, providing an accurate and understandable model for spam accounts identification. This research represents a notable progress in the domain, offering a thorough and dependable resolution for spam accounts detection issue.