Reproduce “Fair Reciprocal Recommendation in Matching Markets” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Yoji Tomita, Tomohiki Yokoyama

The authors develop a reciprocal recommendation method that balances expected mutual matches across both sides. Envy measures whether a user values another same-side user’s recommendation opportunity more than their own. In simulations, their Nash social welfare method sharply reduces envy while retaining competitive match counts, including with sampled Japanese-platform preference estimates; these are offline results, not a live-platform trial.

Key insights

  • An empirical trade-off exists between maximizing the total expected number of matches (social welfare) and ensuring fairness of recommendation opportunities in RRSs.
  • Maximizing Nash Social Welfare (NSW) serves as a principled approach to achieve a balanced distribution of recommendation opportunities, promoting fairness among agents.
  • A theoretical guarantee shows that NSW-best recommendations can achieve approximate double envy-freeness under certain conditions, particularly for single-position recommendations.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.