Uncovering Online Job Application Deception Using Deep Learning Frameworks

In recent years, most businesses have shifted to digital platforms for recruitment, streamlining the hiring process but also increasing the risk of fraudulent job postings.Scam artists are using these platforms to create fraudulent listings for their own profit, which when leveraged for these purposes is real risk for online recruitment fraud as a subset of cybercrime.This research project explored the possibility of detecting these scams through the use of two transformer-based deep learning models BERT and RoBERTa.The study produced a new dataset with three data sources and also included jobs posted very recently that could resolve some of the outdated nature of the benchmark datasets.Exploratory Data Analysis (EDA) revealed a considerable class imbalance that could introduce bias against the minority classes and information loss.To deal with the considerable class imbalance issue and potential bias we pursued and compared ten variations of SMOTE