Detecting Fake Job Postings Using BI-LSTM and Attention Mechanism

Abstract

With the advent of online medias, fakeness in job have been increased. The rapid growth of job sites that offering numerous jobs multiplied the amount of fakeness. Through fraudulent job postings many innocent people have huge financial losses and which in turn leads to identity theft also. Therefore we have to devise effective methodologies to avoid its impact. Various machine learning and Natural Language Processing techniques have been used in various domains such as fake news detection, e-mail spam detection etc. Using Recurrent Neural Networks and its variants can capture the patterns in the text. In this paper a new idea is presenting in which fake jobs notification in online medias can be detected with BI-LSTM and attention mechanism. Both methods can capture the dependencies between text more efficiently and helps to detect the fakeness in job sites more accurately.

Introduction

In the digital era, online job postings have become widespread, but the rise of fake job advertisements poses risks to job seekers. This paper proposes a robust detection system using Bi-directional Long Short-Term Memory (Bi-LSTM) with an attention mechanism to identify fraudulent job postings. Bi-LSTM captures contextual information from both past and future words in job descriptions, while the attention mechanism focuses on critical features such as suspicious phrases, abnormal requirements, and key textual cues. The model also incorporates structured data like job location, type, and salary to improve detection accuracy.

Methodology:

1. Preprocessing: Tokenization, cleaning, and padding of text data.

2. Model Design: Bi-LSTM layers for sequential context, attention layers for highlighting important features, followed by dense and output layers for classification.

3. Training & Evaluation: Model trained on datasets with hyperparameters like embedding dimension 100, 128 LSTM units, batch size 25, and sequence length 100–150 tokens.

Results:

• The Bi-LSTM with attention mechanism achieved 99.38% accuracy, outperforming prior models.

• High performance metrics: Precision 0.98, Recall 0.99, F1-score 0.99.

• The model effectively captures long-range dependencies and enhances interpretability by highlighting influential parts of job postings.

This approach addresses previous research gaps by combining deep learning, attention mechanisms, and both structured and unstructured data, creating a scalable, accurate, and interpretable solution for detecting fake job postings in real time.

Conclusion

In this article we presented an approach to detect fake job posting using BILSTM and attention mechanism and from the experimental results we got this model acquired 99% accuracy. So from this study, we can conclude that this model can detect fake jobs very efficiently. Therefore this model is a boon for job seekers from getting trap into fake job sites and they can make them safe from themselves from financial and mental tortures.

References

[1] K. Sridevi, G. Likitha , Real or Fake Job posting detection,2024 [2] Maddi Sravya Reddy, Maddikera Hemanth Lal, Fake Job post detection using machine learning ,2024. [3] V Anbarasu, Dr. S. Selvakani, Mrs. K. Vasumathi, Fake Job Prediction Using Machine Learning,2024. [4] Sherina Sara Jaison, Mallikarjuna Kodabagi , Identifying real and fake job posting using machine learning ,2023. [5] Mrinal Kumari, NSK Satya kala, Nandini R, Fake job posting prediction using machine learning approach ,2023. [6] Hina Afsal,Furqan Rustam,Identifying fake job posting using selective features and resampling techniques,2023. [7] Aravind Sasidharan Pillai,Detecting fake job posting using Bidirectional LSTM,2023.

Copyright

Copyright © 2026 Shilpa K, Renjith M. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.