Fairness of Exposure in Rankings

By Ashudeep Singh, Thorsten Joachims. 2018

692 citations recorded by the source corpus.

Rankings are ubiquitous in the online world today. As we have transitioned from finding books in libraries to ranking products, jobs, job applicants, opinions and potential romantic partners, there is a substantial precedent that ranking systems have a responsibility not only to their users but also to the items being ranked. To address these often conflicting responsibilities, we propose a conceptual and computational framework that allows the formulation of fairness constraints on rankings in terms of exposure allocation. As part of this framework, we develop efficient algorithms for finding rankings that maximize the utility for the user while provably satisfying a specifiable notion of fairness. Since fairness goals can be application specific, we show how a broad range of fairness constraints can be implemented using our framework, including forms of demographic parity, disparate treatment, and disparate impact constraints. We illustrate the effect of these constraints by providing empirical results on two ranking problems.

Subfields: Economics and Econometrics, Management Science and Operations Research

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Economics and Econometrics

In Management Science and Operations Research

Citations in the corpus

FairSight: Visual Analytics for Fairness in Decision Making (2019, background): “The fairness problem here is to pursue the degree of exposure and attention fairly for groups [37, 39] or individuals [4].”
Investigating Societal Biases in a Poetry Composition System (2020, background): “More recently, researchers have proposed different metrics to measure the amount of group and individual fairness in ranking systems (Zehlike et al., 2017; Biega et al., 2018; Singh and Joachims, 2018; Yang et al., 2020).”
How Can Recommender Systems Benefit from Large Language Models: A Survey (2023, neutral): “The item-side fairness in recommender systems ensures that each item or item group receives a fair chance to be recommended ( e.g. , proportional to its merits or utility) [Patro et al. , 2020; Liu et al. , 2019; Singh and Joachims, 2018].”
Using Image Fairness Representations in Diversity-Based Re-ranking for Recommendations (2018, background): “Similarly, Singh and Joachims [16] proposed a framework that adds fairness constraints on rankings.”
Fairness in Recommendation Ranking through Pairwise Comparisons (2019, background): “r dilemma. Recent research developed fairness metrics centered around pointwise accuracy [8, 49], but this does not indicate much about the resulting ranking that the user actually sees. In contract, [52, 44, 45, 11] explore what is a fair ranking, but focus on unpersonalized rankings where relevanc”
Equity of Attention: Amortizing Individual Fairness in Rankings (2018, methodology): “This goal is achieved by techniques similar to those for ranking result diversification [6, 34, 36], or by granting equal ranking exposure to groups [30].”
Fairness-Aware Ranking in Search & Recommendation Systems with Application to LinkedIn Talent Search (2019, result): “Our work is closely related to recent literature on fairness in ranking [5, 7, 12, 14, 36, 39, 41].”
Optimizing Long-term Social Welfare in Recommender Systems: A Constrained Matching Approach (2020, background): “Some work has looked at the impact of RS policies on provider welfare (Singh & Joachims, 2018), on long-term user and provider metrics (e.g., using RL (Chen et al., 2018; Ie et al., 2019b) or assessing various other phenomena (Ribeiro et al., 2020; Celma, 2010)).”
A Survey on the Fairness of Recommender Systems (2022, methodology): “Divided by Optimization Object Treatment [20, 29, 30, 33, 36, 48, 55, 60, 76, 83, 93, 95, 97, 103, 115] [19, 28, 31, 45, 58, 59, 62, 67, 71, 82, 86, 90, 107, 112, 116] [7, 8, 12, 15, 23, 44, 52, 64, 75, 81, 84, 85, 87, 94, 114] [61, 66, 77, 80, 99, 102, 106] Impact [6, 19, 22, 25, 33, 39, 43, 53, 54”
Sociotechnical Harms: Scoping a Taxonomy for Harm Reduction (2022, background): “These harms occur when a system withholds information, opportunities, or resources [22] from historically marginalized groups in domains that aﬀect material well-being [127], such as housing [41], employment [176], social services [15, 176], ﬁnance [100], education [102], and healthcare [139].”
Towards a Fair Marketplace: Counterfactual Evaluation of the trade-off between Relevance, Fairness & Satisfaction in Recommendation Systems (2018, background): “Recent work has explored the development of classification models with fairness-aware regularization [30], fairness aware decision making systems [9], prediction models for individual fairness [6], fair division of resources [2], auditing for fairness [17] and fairness in rankings [4, 25].”
Towards Intersectionality in Machine Learning: Including More Identities, Handling Underrepresentation, and Performing Evaluation (2022, background): “13We note here a difference in what we are proposing from the syntactically similar space of fairness in rankings [114, 123] There, those being ranked are individuals, and the goal is to more closely align to a set of ground-truth rankings.”
Overcoming Failures of Imagination in AI Infused System Development and Deployment (2020, background): “For instance, while restricted access to some critical service for some stakeholders (Abbasi et al., 2019) affects them directly, the exposure or representation of stakeholders (Biega et al., 2018; Singh and Joachims, 2018) may also affect those creating or consuming the content.”
Fairness in Machine Learning: A Survey (2020, background): “…strategies for a variety of (fairness) metrics [151], subset-based evaluation metrics to measure the utility of recommendations for different groups (e.g., based on demographics) [93], and a general framework to optimize utility metrics under fairness of exposure constraints [264, 265].”
Multistakeholder recommendation: Survey and research directions (2020, neutral): “5.5.1belowaswellas recentworkinfairnessinranking(Zehlikeetal.2017a,b;SinghandJoachims2018; Beuteletal.2019).”
User-oriented Fairness in Recommendation (2021, background): “As concluded in [32], the metrics include the supervised criteria which control the average exposure of groups to be proportional to the average relevance of the results of groups to a query [8, 39], and the unsupervised criteria which require that the average exposure at the top of the ranking list”
Sociotechnical Harms of Algorithmic Systems: Scoping a Taxonomy for Harm Reduction (2022, background): “In the employment domain, recommender or ranking systems that match employers and potential candidates may prioritize the resumes of men over other genders [201, 224].”
A Survey of Graph Neural Networks for Recommender Systems: Challenges, Methods, and Directions (2021, methodology): “Speciically, researchers adopt two methods to enhance fairness: One is directly debiasing recommendation results in the training process [9, 254], while the other endeavors to rank items to alleviate the unfairness in a post-processing method [55, 153].”
Pairwise Fairness for Ranking and Regression (2019, methodology): “Of these, some provide post-processing algorithms for re-ranking a given ranking [Biega et al., 2018, Celis et al., 2018, Singh and Joachims, 2018, 2019], while others, like us, learn a ranking model from scratch [Zehlike and Castillo, 2018, Singh and Joachims, 2019].”