Explainable and Transparent Ranking Functions


Rankings are commonly used to make decisions and allocate resources in a wide variety of applications such as  school admissions, job applications, public housing allocation, sport competition judging, organ donation lists.

This project focuses on accountability, transparency, and explainability in the context of ranking-based decision processes. We focus on understanding the impact of the ranking process a priori, based on the ranking functions and data distribution, to help decision-makers understand the behavior of their ranking functions, and to provide entities being ranked with some transparent and understandable explanation of the ranking process.

This project aims at providing:

  1. Transparency through the design of metrics to assist in making the ranking process clear to both the decision-makers and the entities being ranked, by assessing the expected importance of each parameter used in the ranking process in the creation of the final ranked outcome.
  2. Explainability through the design of human-understandable ranking functions that take into account real-life constraints (e.g., fairness requirements, bounds on the use of some parameters) and that can be shared with non-technical audiences so they know what to expect in the ranking process.
  3. Tools to help decision-makers create transparent and explainable ranking functions designed for their applications.