Abstract

As information access systems have profound impact through their mediation of users’ experiences of information spaces, it is vital to ensure that their benefits, costs, and other impacts are fair: that people have equitable access to information, relevant information results, and opportunities, and are not systematically excluded, underserved, or harmed. Fairness is further a broad and complex topic with many divergent and sometimes competing definitions. This tutorial will provide an introduction to fairness in machine learning and information access, a broad survey of the landscape of fairness and fairness-related harms, and a review of key results and both established and emerging techniques for measuring and improving the fairness of information retrieval technologies.

Key Resources

This tutorial reuses content from the 2019 tutorials authored by myself, Robin Burke, and Fernando Diaz, and much of the material is in our monograph:

Slides

Bibliography