Mauritz Cartier van Dissel presents 'Fairness in rankings through random sampling' in IceLab September 22nd 11.00, in IceLab and in Zoom.
Rankings play a central role in decision-making, from search engine results to hiring practices. But what does it mean for a ranking to be fair? Existing fairness frameworks often emphasize proportional representation of protected groups, yet they tend to ignore the randomness inherent in ranking processes and the practical limitations of finite candidate pools. In this talk, I discuss hyperFA*IR, a framework for evaluating and enforcing fairness in rankings using the hypergeometric distribution. Unlike traditional probabilistic approaches such as binomial models, this method reflects the finite and constrained nature of real-world candidate pools.
Drop in to IceLab (find using mazemap) or join via Zoom: IceLab Living Room Zoom 612 568 7586
Guest researcher Mauritz Cartier van Dissel is a PhD candidate at the Complexity Science Hub in Austria, working with IceLab alumnus Professor Fariba Karimi of the Network Inequality and Fairness Group. He is currently analyzing the emergence of bias in multi-attributed networks and the related possible mitigation techniques. He is also part of the EU-funded MAMMOth project, which aims to promote multi-dimensional fairness in AI systems.
Mauritz is visiting IceLab for a day before taking part in IceLab Camp.
Mauritz Cartier van Dissel, PhD candidate at the Complexity Science Hub in Austria, Network Inequality and Fairness Group