Online Recruitment Fraud (ORF) Detection Using Deep Learning Approaches
DOI:
https://doi.org/10.51483/IJAIML.2915Keywords:
Class imbalance, data augmentation, deep learning, employment scam, fraud detection, machine learning, online recruitment, SMOTE, transformer-based models”.Abstract
The new employment process in the web has revolutionized, but it also has its drawbacks like creating fake job offers and causing financial loss to job seekers. To overcome this challenge, a DL approach is proposed for ORF detection using a new dataset that was gathered from Fake Job Posting, Pakistan Job Posting and US Job Posting datasets. The technique uses BERT and RoBERTa to convert the task details into numeric vectors. To deal with the large class imbalance in the dataset, the SMOTE variation SMOBD is used to balance the classes well. This updated framework integrates these advances with a two-dimensional CNN2D for job classification. The results reveal that the model of BERT model features and SMOBD with CNN2D has the highest classification accuracy of 98.68%. This approach addresses some shortcomings of legacy datasets and gives them a solid response to the challenge of detecting a fake job posting, thereby making a significant contribution towards preventing online recruiting fraud.
