Author

Jatin Unecha

Publication Date

Fall 2024

Degree Type

Master's Project

Degree Name

Master of Science in Computer Science (MSCS)

Department

Computer Science

First Advisor

Teng Moh

Second Advisor

Melody Moh

Third Advisor

Albert Tsao

Keywords

Gale-Shapley algorithm, job recommendation system, stable matching algorithm

Abstract

In today’s competitive job market, recommendation systems are essential for con- necting job seekers with suitable opportunities. However, traditional recommendation

models, such as collaborative and content-based filtering face challenges in ensuring diversity and fairness. This results in low diversity and high congestion around popular job listings which refers to the problem where a few jobs are recommended to large numbers of users as well as imbalanced job exposure. These drawbacks lower the satisfaction for job applicants and reduce the employers’ ability to reach a broader pool of candidates. In this research we address these challenges by introducing the Gale-Shapley algorithm or stable matching algorithm into our job recommendation

system. Our approach focuses on balancing job exposure and aligning the prefer- ences of job seekers and employers more effectively. Our method improves upon

the conventional models by not only reducing congestion in high-demand roles but also increasing diversity and fairness. Through our experiments on a job application dataset, we demonstrate that the stable matching approach improves metrics such as congestion, Gini index, and coverage by a factor of 2 consistently across job domains and varying levels of user-job interactions. We achieve this without compromising accuracy and precision. This shows that the stable matching approach reduces job application competition and supports a more balanced outcome for job seekers. Our results underscore the potential of stable matching-based systems to enhance the fairness and effectiveness of job recommendation systems.

Available for download on Friday, January 23, 2026

Share

COinS