A machine learning approach to detecting fraudulent job types

Abstract Job seekers find themselves increasingly duped and misled by fraudulent job advertisements, posing a threat to their privacy, security well-being. There is clear need for solutions that can protect innocent seekers. Existing approaches detecting jobs do not scale well, function like black-box, lack interpretability, which essential guide applicants’ decision-making. Moreover, commonly used lexical features may be insufficient as the representation does capture contextual semantics of underlying document. Hence, this paper explores what extent different categorizations classified. In addition, seeks type are most relevant in classifying job. paper, we develop validate machine learning system identifying identity theft, corporate theft multi-level marketing amongst advertisements. We utilized four classes features: empirical rule set-based features, bag-of-word models, recent state-of-the-art word embeddings transformer models various classifiers. The were validated evaluating them on publicly available description dataset. Our results indicate transformer-based consistently outperformed handcrafted rule-set based class. Ultimately, Gradient Boosting classifier with combination parts-of-speech tags bag-of-words vectors achieved best performance an F1-score 0.88.