The Shapley value is a very useful concept to evaluate the importance/contribution of a player based on how he affects different possible coalitions. Now based on this information, is it possible to build a network between players that tells us, for each player $p_i$, which relationships can yield more/less successful outcomes ?

To be clear, let's say that we have a cooperative game with three players : $N=\{p_1,p_2,p_3\}$ and an evaluation function $f$ defined over the power set of players.

  1. I want to know how important each player is with respect to the group (easily done with the Shapley vlue).
  2. I want to be able to draw conclusions that relate each player to the others, something like : Player $p_1$ yields better outcome when cooperating with $p_2$ than with $p_3$, and $p_3$ is better off playing by himself.

Is there any concept that can help evaluate such relashionships or maybe build a whole "social network" between players ?

Thanks !

  • $\begingroup$ I feel that this should probably be more application-specific. It depends on what you want the edges in this graph to model (in my opinion). I would also guess that aspects of this question have been studied in economics, but I'm not sure where to start looking for references.... $\endgroup$
    – usul
    Commented May 23, 2016 at 14:42


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.