DS Dissertation Proposal | Weixiao Huang | Thursday, April 18, 4:00PM - 5:00PM | Gordon Library 303 Conference Room

Data Science 

Ph.D. Dissertation Proposal  

Weixiao Huang, Ph.D. Candidate 

Thursday, April 18, 2024

4:00PM - 5:00PM

Gordon Library 303 Conference Room 

 Zoom link available upon request - contact datasciecne@wpi.edu

Dissertation Committee:

Dr. Andrew Trapp, Advisor, Worcester Polytechnic Institute

(Below in alphabetical order)

Dr. Alexander Teytelboym, University of Oxford

Dr. Jennifer Pazour, Rensselaer Polytechnic Institute

Dr. Oren Mangoubi,  Worcester Polytechnic Institute

Title:  Mathematical Optimization and Market Design for Resource Sharing in the Nonprofit Sector

Abstract:

Resource allocation and resource matching are ubiquitous in a wide variety of domains, from private, public, to non-profit sectors. Market design and mathematical optimization produce state-of-art allocating and matching results for these applications. In this dissertation, I address resource sharing among nonprofit organizations (NPOs) through two main directions: 1. designing a system for overall beneficial and fair exchanges of resources to empower collective impact, and 2. developing new ways to form cohorts of NPOs to share resources while ensuring organizational compatibility. 

Nonprofits lack resources, hindering the quality and quantity of service they can deliver. Meanwhile, NPOs at times have underutilized or even spare resources due to the inability to scale expertise in staffing and tangible resources to meet temporally shifting service demands. Motivated by these observations, I propose a novel resource sharing system, SWAP, which to the best of my knowledge, is the first resource sharing system that facilitates resource exchanges where organizations can obtain resources by offering their own. SWAP consists of four elements: a collaborative auction-based sharing process, complete with an offering mechanism, a bidding mechanism, the virtual currency, SWAPCredit, to facilitate liquidity in exchange, and the SWAPCredit (re)-distribution mechanism; a central technology that represents the award determination problem with a multilateral exchange optimization model, generating optimal resource exchange outcomes by accounting for both overall utility and individual fairness; an online platform, the SWAP Hub, where NPOs can easily offer and bid on available resources, and receive exchange results; and human-centric co-design, shaping the understanding and design decisions of researchers and stakeholders. As the SWAP system has been deployed in the real world and more NPOs are interested in joining, the question of how to group NPOs into cohorts or recommend an NPO to join which cohort to ensure positive resource exchange experiences becomes interesting. To solve the problem, I identify attributes that may significantly affect resource sharing, design a data collection method to obtain the necessary data while valuing NPO directors’ time (i.e., to request information by limiting communication with stakeholders), propose and refine an integer program to generate formation outcomes, and conduct experiments with both real and simulated data to demonstrate the model.