About the project
This project considers the problem of two-sided matching markets where two distinct sets of agents need to be matched with each other. This problem has a wide range of applications, e.g., in labour markets such as online crowd sourcing platforms (e.g., Amazon Mechanical Turk), ride sharing or online dating services. Agents in the market have preferences over the agents on the other side of the market while they compete with each other for the final match. An additional challenge in the online setting is that agents are usually uncertain about their preferences, and they often acquire knowledge of their preferences from temporary matches. When matchings are facilitated by a central mechanism, it is usually desired for the matching to be fair w.r.t. the agents' preferences. We investigate how a stable matching can be learned efficiently over time. Here, stability is a notion that imposes that no two agents would prefer to be matched to each other compared to the partner that was suggested by the centralised mechanism.
The objective in this project is to consider sampling strategies that allow for fast identification of a stable matching. In particular we adjust common sampling strategies from the Multiarmed-Bandits literature to this setting. As a starting point, we consider the case where only one side of the market has uncertain preferences.
This project is a collaboration with external collaborators from University of Neuchatel, Switzerland.
Project team
- Postdoctoral Fellow Anne-Marie George, University of Oslo, Department of Informatics
- Professor Christos Dimitrakakis, University of Oslo, Chalmers University, University of Neuchatel
- PhD Candidate Andreas Athanasopoulos, University of Neuchatel