Master Thesis Defense - August 6, 2015: Two-sided Matching Algorithm for Dynamic Labor Markets

Xinkai Xu

Thursday, August 6, 2015 at 11:00 a.m.
Room EV011.119

You are invited to attend the following M.A.Sc. (Quality Systems Engineering) thesis examination.

Examining Committee

Dr. A. Awasthi, Chair
Dr. Y. Zeng, Supervisor
Dr. C. Wang, Co-Supervisor
Dr. A. Ben Hamza, CIISE Examiner
Dr. H. Ge, External Examiner (BCEE)

Abstract

Temporary recruiting is a fast growing labor market with candidates, usually professionals, looking for contract positions in today;s fast evolving and dynamic economy.  While two-sided matching algorithms have been applied to some labor markets during the past decades, they usually assume a static environment in which the strategic interactions between candidates and employers are designed without considering dynamic changes from market participants.  In recent years, the advance of Internet and mobile technologies has enabled the implementation of large scale online labor markets which can capture the dynamics of the participants in a near real time manner.

In this thesis, I study how to improve effectiveness and efficiency of the recruitment process in the context of temporary labor markets.  The key challenge is how to design two-sided matching algorithms to accommodate dynamic changes from market participants.  To have a better understanding of the challenge and possible solutions, I first analyzed the problem using the Environment-Based Design (EBD) methodology.  The analyzing results point to two important requirements to be addressed in the proposed two-sided matching algorithm: efficiency, which is the responsiveness of the algorithm, and effectiveness, which is measured as the consistency of solutions generated by the algorithm when dynamics are introduced.

Based on the results derived from EBD analysis, I designed two algorithms to address the requirements of efficiency and effectiveness.  To improve the efficiency of two-sided matching, I designed a request-accumulated deferred acceptance algorithm, which is a modification of the classical deferred acceptance algorithm.  In addition, I designed a repair-based two-sided matching algorithm to improve solution consistency when changes are introduced to the market.  The performance of the proposed algorithms is evaluated through a computational study.  Results show that the efficiency an effectiveness design requirements are satisfied.