GUARANTEED USER FAIRNESS IN RECOMMENDATION
Keywords: Recommendation Systems, Fairness in RS, Conformal Prediction
Abstract: Although recommender systems (RS) have been well-developed for various fields of applications,
they suffer from the crisis of platform credibility with respect to RS confidence and fairness, which
may drive users away from the platform and result in the failure of the platform’s long-term success.
In recent years, a few works have tried to solve either the model confidence or fairness issue,
while there is no statistical guarantee for these methods. It is therefore an urgent need to solve
both issues with a unifying framework with statistical guarantee. In this paper, we propose a novel
and reliable framework called Guaranteed User Fairness in Recommendation (GUFR) to dynamically
generate prediction sets for users across various groups, which are guaranteed 1) to include
the ground-truth items with user-predefined high confidence/probability (e.g., 90%); 2) to ensure
user fairness across different groups; 3) to have the minimum average set size. We further design an
efficient algorithm named Guaranteed User Fairness Algorithm (GUFA) to optimize the proposed
method, and upper bounds of the risk and fairness metric are derived to help speed up the optimization
process. Moreover, we provide rigorous theoretical analysis with respect to risk and fairness
control as well as the minimum set size. Extensive experiments also validate the effectiveness of the
proposed framework, which aligns with our theoretical analysis. The code is publicly available at
https://anonymous.4open.science/r/GUFR-76EC.
Primary Area: alignment, fairness, safety, privacy, and societal considerations
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 1618
Loading