Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
February 25, 2026ACM Transactions on Recommender SystemsOpen Access

Balancing Fairness and High Match Rates in Reciprocal Recommender Systems: A Nash Social Welfare Approach

View Full Paper
Ask AI
Bookmark
Share

Authors

YTYoji TomitaCyberAgent (Japan)TYTomohiro YokoyamaShibaura Institute of Technology

Discussion

Loading...

Member takes

Implication

This research demonstrates balancing fairness and match rates in reciprocal recommender systems, indicating potential for equitable user opportunities in matching platforms.

Key Points

  • The central aim is to enhance matching in reciprocal recommender systems while ensuring fairness among users.
  • Introduce the Social Welfare method to approximate match maximization.
  • Define and implement envy-freeness for fair allocation of recommendations.
  • Propose the Nash Social Welfare method to optimize fairness and match rates.
  • Generalize the methods to the α -SW framework for balanced outcomes.
  • Develop a computationally efficient approximation algorithm using the Sinkhorn algorithm.
  • The Social Welfare method significantly increases matches but leads to unfairness.
  • The Nash Social Welfare method achieves nearly envy-free recommendations.
  • The α -SW method effectively balances trade-offs between fairness and match rates in practical scenarios.

Cite This Study

Tomita et al. (2026) studied this question.

synapsesocial.com/papers/699e912ef5123be5ed04e80fhttps://doi.org/10.1145/3793538
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 4 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1The Combinatorial Assignment Problem: Approximate Competitive Equilibrium from Equal Incomes2011 · 801 citations
  2. 2Consensus of Subjective Probabilities: The Pari-Mutuel Method1959 · 430 citations
  3. 3An Empirical Study of Rich Subgroup Fairness for Machine Learning2019 · 198 citations
  4. 4Optimizing Rankings for Recommendation in Matching Markets2022 · 19 citations