Comparative Analysis of Fake Job Posting Detection Using Natural Language Processing and Machine Learning

This paper has presented a comprehensive comparative analysis of seven machine learning and deep learning classifiers for the detection of fraudulent online job postings. Utilizing the EMSCAD benchmark dataset and an end-to-end NLP preprocessing pipeline incorporating TF-IDF vectorization, the proposed study systematically evaluates KNN, Decision Tree, Naive Bayes, SVM, Random Forest, MLP, and DNN under identical experimental conditions. The experimental evidence unequivocally establishes the Deep Neural Network as the superior classifier for this task, achieving 98.0% accuracy, 95.7% precision, 93.4% recall, and an F1-score of 94.54% — all substantially outperforming traditional machine learning baselines. The DNN's capacity for hierarchical feature representation enables it to capture complex non-linear relationships within the high-dimensional TF-IDF feature space that shallower models cannot adequately model. The practical deployability of the proposed system is demonstrated through a functional Flask web application that provides real-time fake job post classification via both structured feature input and free-text job description analysis. This deployment bridges the critical gap between research-stage models and operational fraud detection tools suitable for integration into live recruitment platforms. The study contributes three principal advances to the employment fraud detection literature: (1) the most comprehensive seven-algorithm comparative benchmark reported to date under uniform experimental conditions; (2) a validated hybrid feature engineering strategy combining NLP-derived TF-IDF features with structured categorical attributes; and (3) a fully deployable, open-architecture web application framework for real-time fraud detection. These contributions collectively advance the field toward practical automated screening solutions that can meaningfully reduce the prevalence of employment scams and protect vulnerable job seekers in the digital economy.

[1] S. Vidros, C. Kolias, G. Kambourakis, and L. Akoglu, "Automatic detection of online recruitment frauds: Characteristics, methods and future directions," Future Internet, vol. 9, no. 6, pp. 1-17, 2017. [2] A. S. Pillai, "Detecting fake job postings using Bidirectional LSTM," arXiv preprint arXiv:2304.02019, 2023. [3] C. Srikanth, M. Rashmi, S. Ramu, and R. M. R. Guddeti, "A novel fake job posting detection: Performance evaluation using ML and ensemble techniques," Lecture Notes in Electrical Engineering, vol. 1049, pp. 219-234, 2023. [4] A. Ranga Chakradhar, V. Amutha Mary, R. Avinash, Jesudoss, and S. P. Shyry, "A comparative study on fake job post prediction using different machine learning techniques," Mathematical Statistician and Engineering Applications, vol. 71, no. 3, 2022. [5] M. Alghamdi and H. Haddadi, "Employment scam detection using machine learning and text mining," Int. J. Advanced Computer Science and Applications, vol. 12, no. 8, 2021. [6] I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. Cambridge, MA: MIT Press, 2016. [7] J. Devlin, M. Chang, K. Lee, and K. Toutanova, "BERT: Pre-training of deep bidirectional transformers for language understanding," in Proc. NAACL-HLT, Minneapolis, MN, 2019, pp. 4171-4186. [8] A. Vaswani et al., "Attention is all you need," in Advances in Neural Information Processing Systems (NeurIPS), vol. 30, 2017, pp. 5998-6008. [9] T. Mikolov, K. Chen, G. Corrado, and J. Dean, "Efficient estimation of word representations in vector space," in Proc. ICLR, 2013, pp. 1-12. [10] L. Breiman, "Random forests," Machine Learning, vol. 45, no. 1, pp. 5-32, 2001.