Detecting Fraudulent Job Postings: An Analysis of Machine Learning Approaches

Abstract: The rise of online job platforms has greatly simplified the global job search process for millions of individuals. However, this convenience has also led to a concerning issue - an increase in fraudulent job postings. Job seekers are becoming more susceptible to scams, identity theft, and financial fraud as malicious entities post deceptive job advertisements. This paper conducts a thorough analysis of machine learning approaches aimed at detecting fraudulent job postings, with the main goal of improving the legitimacy and safety of online job markets. The study encompasses data collection, preprocessing, feature engineering, model selection, and evaluation. To tackle these challenges, we explore the effectiveness of various machine learning models, including Decision Trees, Random Forests, and Support Vector Machines (SVM). Additionally, we investigate data preprocessing techniques such as label encoding and standardization to prepare the data for modeling. The results suggest that while Decision Trees and Random Forests show promising performance, dealing with the imbalanced nature of the dataset requires the application of oversampling techniques for enhanced accuracy in identifying fraudulent postings. The paper provides a comprehensive evaluation of model performance, utilizing metrics like accuracy, F1-score, precision, and recall. Furthermore, it emphasizes the significance of strategies to handle class imbalance in real-world applications