Show simple item record

dc.contributor.authorSingh, Ashudeep
dc.date.accessioned2021-12-20T20:48:51Z
dc.date.available2021-12-20T20:48:51Z
dc.date.issued2021-08
dc.identifier.otherSingh_cornellgrad_0058F_12577
dc.identifier.otherhttp://dissertations.umi.com/cornellgrad:12577
dc.identifier.urihttps://hdl.handle.net/1813/110649
dc.description186 pages
dc.description.abstractRanking-based interfaces are ubiquitous in today's multi-sided online economies (such as online marketplaces, job search, property renting, media streaming). In these systems, the items to be ranked are products, job candidates, creative content, or other entities that transfer economic benefit. It is widely recognized that the position of an item in the ranking has a crucial influence on its exposure which directly translates into economic opportunity. Surprisingly, learning-to-rank (LTR) approaches typically do not consider their impact on the opportunity they provide to the items. Instead, most LTR algorithms solely focus on maximizing the utility of the rankings to the user issuing the query, while there is evidence that this does not necessarily lead to rankings that would be considered fair or desirable in many situations. This thesis proposes a conceptual and computational framework that allows the formulation of fairness constraints on rankings in terms of a merit-based exposure allocation. As a part of this framework, we develop efficient learning-to-rank algorithms that maximize the utility for the user while provably satisfying a specifiable notion of fairness. Since fairness goals can be application-specific, we show that a broad range of fairness constraints can be implemented in this framework using its expressive power to link relevance, merit, exposure, and impact. Beyond the theoretical evidence in deriving the frameworks and algorithms, empirical results on simulated and real-world datasets verify the effectiveness of the approach on both individual and group-fairness notions.
dc.language.isoen
dc.rightsAttribution 4.0 International
dc.rights.urihttps://creativecommons.org/licenses/by/4.0/
dc.subjectfairness
dc.subjectmachine learning
dc.subjectranking
dc.titleFairness of Exposure for Ranking Systems
dc.typedissertation or thesis
thesis.degree.disciplineComputer Science
thesis.degree.grantorCornell University
thesis.degree.levelDoctor of Philosophy
thesis.degree.namePh. D., Computer Science
dc.contributor.chairJoachims, Thorsten
dc.contributor.committeeMemberSridharan, Karthik
dc.contributor.committeeMemberMimno, David
dc.contributor.committeeMemberBarocas, Solon Isaac
dcterms.licensehttps://hdl.handle.net/1813/59810
dc.identifier.doihttps://doi.org/10.7298/t38f-gx71


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record

Except where otherwise noted, this item's license is described as Attribution 4.0 International

Statistics