Recognition Of Fraudulent Job Advertisements Using A Machine Learning Framework

Abstract

The Rapid Growth Of Online Job Markets Has Led To A Rise In Fraudulent Postings, Causing Financial And Emotional Harm To Job Seekers. This Study Introduces An Intelligent Fraud Detection System Using Ensemble Machine Learning And NLP To Automatically Identify Deceptive Job Listings. The Model Utilizes 33 Engineered Features From Text, Structure, And Metadata, Combined Through Random Forest, Logistic Regression, SVM, And Naive Bayes With A Weighted Voting Mechanism, Achieving 95.2% Accuracy, 92% Precision, And 89% Recall On 17,880 Verified Postings. Integrating External Company Verification, Domain Trust Checks, And Blacklist Monitoring Enhances Detection Confidence.. Results Outperform Single Models And Rule-based Systems, Offering Both Practical And Theoretical Advancements In Online Job Fraud Prevention. Experimental Results Demonstrate Significant Performance Improvements Over Single-algorithm Approaches And Traditional Rule-based Systems, Particularly In Detecting Sophisticated Fraud Patterns That Evade Conventional Detection Methods.

Keywords