A game theoretic approach to value information in data mining

This paper applies a game-theoretic framework to suggest a fair value for information extracted via data raining and shared between two retail market competitor firms. Neither firm has a dominant position in that market. Two players, each owning a privileged information set (a collection of data) may wish to share or pool that information for mutual benefit. We assume that each player is equipped with a mining technique which extracts information from the data. We first model information sharing as a cooperative game. Then we use results from the cost sharing literature to construct information sharing methods when data can be quantified either as discrete or as continuous variables. In the latter case, we supply a method to obtain decision rules for price shared information.