Predicting Fake Job Posts with a Voting Classifier of Multiple Classification Models

Abstract

The detection of fake job posts is becoming increasingly important in the modern job market. With the rise of online job postings, scammers and fraudulent actors are taking advantage of unsuspecting job seekers by posting fake job listings that appear legitimate. This paper proposes a machine learning approach to detect fake job posts using a combination of textual and categorical data. We extract various features from the job post text, such as the presence of certain keywords, as well as features from the job post, such as the job title, employment type, required experience. Models like Logistic regression, SVM, Decision tree, Random forest, Gradient boosting, XGBoost, and MLP with Adam optimizer are compared using various metrics like accuracy, F1 score, ROC AUC score, and more after training. This research can be used to build automated systems to detect fake job posts, helping to protect job seekers from scams and fraudulent activities in the job market.