Truthful and Stable One-sided Matching on Networks

Tianyi Yang, Yuxiang Zhai, Dengji Zhao, Xinwei Song, Miao Li

Abstract: Diffusion one-sided matching aims at incentivizing more participants to match so as to improve overall matching result. Existing works have tried to add constraints on Top Trading Cycles to obtain the incentive, but it only works in trees. In this paper, we first propose a mechanism named Swap With Neighbors (SWN), which can work in any graph structure and intuitively satisfy incentive compatibility and the tightest stability (first defined here) in the new setting. Then we find a natural improvement of SWN called Leave and Share which not only reaches the same properties as SWN but also provides an obvious efficiency difference.