Detecting Online Recruitment Fraud (ORF) Using Deep Learning Techniques Enhancing With Convolution Neural Networks (CNN 2D)

Abstract

The emergence of online recruitment platforms has revolutionized the hiring process, making it much easier and efficient. However, they have also led to an increase in fraudulent job ads, giving rise to financial and informational hazards for job searchers. This study suggests to overcome this challenge, a deep learning approach for detection of Online Recruitment Fraud (ORF) is proposed based on the combination of data from Fake Job Posting, Pakistan Job Posting, and US Job Posting. This is done using a method that leverages Bidirectional Encoder Representations from Transformers (BERT) and Robustly Optimized BERT Pretraining Approach (RoBERTa) to generate numerical representations of job descriptions within context. The SMOBD approach which is based on the Synthetic Minority Oversampling Technique (SMOTE) is employed to effectively balance data, which is a kind of approach that is used to solve the class imbalance problem. Then the feature representations produced are concatenated with a Two-Dimensional Convolutional Neural Network (CNN2D) for classification. Through the experimental results, it is found that the combination of BERT embeddings, SMOBD and CNN2D achieves the highest classification accuracy of 98.68%. The suggested framework increases the efficacy of fraud detection and provides an effective way to identify fraudulent job posts, contributing to safer and more dependable online recruitment platforms.