Explainable Ensemble Learning with SHAP for Fraudulent Job Posting Detection

This paper presented Fraud Shield PRO, an explainable AI-based framework for detecting fraudulent job postings using machine learning and NLP techniques. The proposed system utilized TF-IDF feature extraction, behavioral analysis, and SMOTE-based class balancing to improve fraud detection performance. A stacked ensemble model combining Logistic Regression, Extra Trees, and Gradient Boosting achieved an accuracy of 98.5% with high classification reliability. SHAP explainability techniques were integrated to provide transparent and interpretable predictions. In addition, a 0–100 risk scoring framework improved fraud assessment and decision-making capability. The system was deployed on Streamlit and Flask for real-time prediction and analysis of fraud.   Limitations: The developed system was trained using only one public dataset, which can affect its performance on various online job portals and in real-life settings. The system only considered text and behavior-based features, while multimedia contents or external links, are not considered. Although the system achieved high accuracy, evolving fraud strategies may affect long-term prediction performance. In addition, lightweight feature engineering techniques were preferred over computationally expensive transformer models to maintain real-time deployment efficiency.   Take-aways: The proposed Fraud Shield PRO system demonstrates that combining ensemble learning, explainable AI, and risk scoring can effectively improve fraudulent job detection accuracy and transparency. The framework also supports scalable real-time deployment for modern online recruitment platforms.

[1] A. Amaar, W. Aljedaani, F. Rustam, V. Rupapara, S. Ludi, and S. Ullah, “Detection of Fake Job Postings by Utilizing Machine Learning and Natural Language Processing Approaches,” Neural Processing Letters, vol. 54, no. 3, pp. 2219–2247, 2022. [2] M. Naudé, K. J. Adebayo, and R. Nanda, “A Machine Learning Approach to Detecting Fraudulent Job Types,” AI & Society, vol. 38, no. 2, pp. 1013–1024, 2023. [3] C. S. Anita, P. Nagarajan, G. A. Sairam, P. Ganesh, and G. Deepakkumar, “Fake Job Detection and Analysis Using Machine Learning and Deep Learning Algorithms,” Revista Gestão Inovação e Tecnologias, vol. 11, no. 2, pp. 642–650, 2021. [4] R. Lal, R. Jaiswal, N. Sardana, A. Verma, A. Kaur, and R. Mourya, “ORFDetector: Ensemble Learning Based Online Recruitment Fraud Detection,” in Proceedings of the International Conference on Contemporary Computing (IC3), 2019, pp. 1–5. [5] S. Dutta and S. K. Bandyopadhyay, “Fake Job Recruitment Detection Using Machine Learning Approach,” International Journal of Engineering Trends and Technology, vol. 68, no. 4, pp. 48–53, 2020. [6] F. Shibly, U. Sharma, and H. Naleer, “Performance Comparison of Two-Class Boosted Decision Tree and Two-Class Decision Forest Algorithms in Predicting Fake Job Postings,” Annals of the Romanian Society for Cell Biology, vol. 25, no. 4, pp. 2462–2472, 2021. [7] H. Afzal, F. Rustam, W. Aljedaani, M. Abubakar, S. Ullah, and I. Ashraf, “Identifying Fake Job Posting Using Selective Features and Resampling Techniques,” Multimedia Tools and Applications, 2023. [8] J. Devlin, M. W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding,” in Proceedings of NAACL-HLT, 2019, pp. 4171–4186. [9] V. Sanh, L. Debut, J. Chaumond, and T. Wolf, “DistilBERT: A Distilled Version of BERT: Smaller, Faster and Cheaper,” 2019. [10] A. Sasidharan Pillai, “Detecting Fake Job Postings Using Bidirectional LSTM,” arXiv preprint arXiv:2304.02019, 2023.