Fernaz Narin Nur, Shafika Rahman, Arafat Sahin Afridi, Nazmun Moon, Shaheena Sultana and A.H.M.Saiful Islam
Adv. Artif. Intell. Mach. Learn., 5 (3):4356-4378
1. Shafika Rahman: Department of CSE, Notre Dame University Bangladesh, Motijheel, 1214, Dhaka, Bangladesh.
2. Arafat Sahin Afridi: Department of CSE, Military Institute of Science and Technology, Mirpur, 1216, Dhaka, Bangladesh.
3. Nazmun Moon: Department of CSE, Daffodil International University, Savar, 1216, Dhaka, Bangladesh.
4. Shaheena Sultana: Department of CSE, Notre Dame University Bangladesh, Motijheel, 1214, Dhaka, Bangladesh.
5. A.H.M.Saiful Islam: Department of CSE, Notre Dame University Bangladesh, Motijheel, 1214, Dhaka, Bangladesh.
6. Fernaz Narin Nur: Military Institute of Science and Technology
DOI: 10.54364/AAIML.2025.53242
Article History: Received on: 08-Jun-25, Accepted on: 10-Sep-25, Published on: 17-Sep-25
Corresponding Author: Fernaz Narin Nur
Email: fernaznur@gmail.com
Citation: Shafika Rahman, Fernaz Narin Nur, Arafat Sahin Afridi, A.H.M.Saiful Islam, Shaheena Sultana, Nazmus Nessa Moon. JoTPaRS: A Job Title Prediction and Recommendation System for IT Professionals. Advances in Artificial Intelligence and Machine Learning. 2025;5(3):242.
In today’s fast-evolving IT landscape, job seekers and recruiters face increasing challenges in aligning candidate qualifications with dynamic job roles and salary expectations. Traditional recruitment systems often rely on manual review or keyword-based algorithms, which lack contextual understanding and fail to provide optimal job recommendations. In this paper, we propose JoTPaRS, a Job Title Prediction and Recommendation System that leverages machine learning and natural language processing to accurately predict relevant IT job titles from user-provided descriptions, skills, and experience levels. In addition to multi-label classification and salary range suggestion, JoTPaRS introduces a novel Game Theory-based optimization layer that selects the most suitable job title among multiple predictions by modeling the decision as a risk-sensitive payoff game. This approach provides users with a ranked optimal job recommendation, enhancing decision support for both job seekers and recruiters. Evaluated on a custom dataset of 8,000 IT job descriptions, JoTPaRS achieved a high prediction accuracy of 95.63\%, outperforming traditional machine learning models and recent deep learning-based benchmarks. User studies further demonstrate increased satisfaction with the system’s recommendations after the inclusion of the Game Theory-based refinement. Our results suggest that combining predictive analytics with decision-theoretic modeling can offer a more precise and personalized job matching experience in the digital hiring ecosystem.