Smart Fraud Detection Framework for Job Recruitments

Abstract

Online recruitment has altered the hiring trend. Specifically, posting job ads on career portals and corporate sites involves seeking a large pool of professional applicants around the globe. Unfortunately, it has been established as another platform for fraudsters, which could lead to loss of privacy for applicants and damages organizations’ reputation. This case study handles the recruitment fraud/scam detection problem. Several important features of organization, job description and type of compensation are proposed and an effective recruitment fraud detection model is constructed using extreme gradient boosting method. It develops an algorithm that extracts required features from job ads and is tested using three examples. The features are further considered for two-step feature selection strategy. The findings show that features of the type of organization are most effective as a stand-alone model. The hybrid composition of selected 13 features demonstrated 97.94% accuracy and outperformed three state-of-the-art baselines. Moreover, the study finds that the most effective indicators are “salary_range,” “company_profile,” “organization_type,” “required education” and “has multiple jobs.” The findings highlight the number of research implications and provide new insights for detecting online recruitment fraud.