Multi-Source Explainable AI for Blockchain Transaction Manipulation Detection Using Supervised and Unsupervised Models

Blockchain has changed the digital finance systems through decentralized, transparent, and tamper-resistant transactions. However, it has led to sophisticated manipulation schemes, like wash trading, rug pulls, and flash loan exploits, due to its open nature. These threats undermine trust and stability in decentralized markets. Machine Learning (ML) models have shown promise in detecting such anomalies. However, ML’s black-box nature poses significant challenges for forensic analysis, user accountability, and regulatory compliance. Hence, to identify and interpret manipulated transaction patterns on public blockchain networks, we propose a novel framework that combines Explainable Artificial Intelligence (XAI) with an ML-based fraud detection scheme. This study proposes a multi-source explainable AI framework for detecting manipulated blockchain transactions using engineered features derived from Ethereum transactional data, scam address blacklists, and phishing datasets. We integrated supervised learning models (XGBoost, LightGBM) and unsupervised approaches (autoencoder neural network, isolation forest) to detect anomalies. We apply SHapley Additive exPlanations (SHAP) to quantify the contribution of each feature in real-time anomaly predictions, offering both global and instance-level explanations. Our model achieves high detection accuracy (F1-score > 0.96) while significantly improving transparency in fraud classification. Furthermore, our visual interpretations with SHAP force plots and transaction graphs permit intuitive auditing of suspicious activities. Our proposed fusion framework of detection and interpretability provides a robust foundation for automated blockchain forensics. This framework helps auditors, developers, and regulators understand the rationale behind flagged anomalies. Our proposed framework advances the development of transparent, scalable, and effective fraud detection systems for decentralized environments.