Evaluating XGBoost and Naive Bayes for Efficient Fraudulent Job Detection: An Explainable Approach
Mohamed Rafi, Mohammed Maruf Hossen, Tazrian Alam · 2025
In recent days there is a significant increase in fraudulent activities because of the dramatic increase of online job postings which creates a most necessary space for the development of robust detection system. This study depicts an accurate approach finding fraudulent job postings using an advanced machine learning techniques by focusing on the application of Natural Language Processing (NLP) and claddification algorithm. We utilized a dataset comprising thousands of job postings, which were meticulously preprocessed using text cleaning, stop-word removal, and TF-IDF vectorization. Classification models, including XGBoost and Gaussian Naive Bayes, were trained and evaluated to determine their effectiveness in identifying fraudulent listings. The results indicated that the XGBoost model outperformed other algorithm, achieving superior accuracy and F1 score. Furthermore, we made use of SHAP (SHapley Additive exPlanations) technique to obtain interpretability, whereby we were able to see the reasoning behind the model’s predictions. In this study, we discuss a new framework for combating online job scams that can be used to protect both the employers and job seekers. The framework is a reliable system that can detect fraudulent job postings and thus job seekers and employers are protected from scams.