Abstract

This thesis designs an automatic two-sided matching system for dynamic labor markets
with large scale of data. Such markets consist of a group of vacancies and applicants, a
matching function, a set of events causing transitions of the state of the market. Due to
the dynamic nature of the labor markets, matching systems based on the classical deferred
acceptance algorithm are not sufficient for producing stable matching solutions. Therefore,
the central theme of this thesis is to address the effectiveness and efficiency of generating
matching results in dynamic large labor markets.
The main contribution of this thesis consists of three dynamic matching algorithms and
a agent-based matching system design. The dynamic matching algorithms are extensions
of the classical deferred acceptance algorithm. The first algorithm generates a vacancyoptimal
stable matching result without considering locking or break-up constrains. The
second algorithm considers locking period constraints in the matching process and the third
algorithm computes applicant-optimal stable matchings with the consideration of break-up
penalties in dynamic environments. To verify the effectiveness and efficiency of the proposed
matching algorithms, theoretical proofs and experimental results are presented as well. The
results indicate that the designed system can be used as an efficient and effective tool for
recruitment management in today’s dynamic and internet based labor markets to reduce
administrative work load of human resource departments and produce stable job allocations.
iii