A Deep Learning-Driven Framework for Identifying Online Recruitment Scams

Abstract

With the expansion of online recruitment platforms, the hiring process has become more efficient; however, this advancement has also led to a rise in fraudulent job postings, causing significant financial harm to job seekers. To address this growing concern, the study introduces a deep learning-based framework for detecting Online Recruitment Fraud (ORF). A novel, comprehensive dataset is constructed by integrating data from Fake Job Postings, Pakistan Job Postings, and US Job Postings. The proposed method leverages state-of-the-art Natural Language Processing (NLP) models—Bidirectional Encoder Representations from Transformers (BERT) and its optimized variant, RoBERTa—to encode job descriptions into meaningful vector representations. To resolve the challenge of class imbalance in the dataset, the Synthetic Minority Over-sampling Technique with Borderline Detection (SMOBD) is employed. These processed features are then input into a two-dimensional Convolutional Neural Network (CNN2D) for classification. Experimental results demonstrate that this integrated approach achieves an impressive classification accuracy of 98.68%. The proposed model not only surpasses existing techniques but also provides a robust and scalable solution for identifying fraudulent job listings, thereby contributing significantly to the prevention of online recruitment scams.