DEEP LEARNING BASED METHODS FOR ONLINE RECRUITMENT FRAUD DETECTION

To make the hiring process simpler, the majority of businesses these days use internet platforms to attract new hires.This has led to misleading advertising as the usage of online job posting sites has grown rapidly.Fraudulent job advertisements are used by scammers to profit from these sites, making online recruitment fraud a major problem in cybercrime.As a result, identifying fraudulent job advertising is crucial to reducing online employment fraud.Recent research have demonstrated the widespread usage of both deep learning and traditional machine learning algorithms to identify fake job advertisements.The goal of this study is to effectively address this problem by utilising Long Short-Term Memory (LSTM) networks.Combining job ads from three distinct sources results in a novel dataset of phoney job postings.Current benchmark datasets are out-of-date and have a narrow scope, which limits how well models work.This constraint is addressed by the inclusion of the most recent job advertisements in the proposed dataset.Exploratory Data Analysis (EDA) draws attention to the issue of class imbalance in the detection of fraudulent employment, which may result in the model's poor performance on minority classes.Ten best-performing Synthetic Minority Oversampling Technique (SMOTE) versions are used in the study to compensate for this.Analysis and comparison are done between the model performances, balanced by each SMOTE version.Out of all the methods used, the LSTM model performed the best in identifying fraudulent job advertisements, achieving an impressive 97% accuracy rate.