An Agent-Based, Multilevel Welfare Assessment Model for Encompassing Assignment and Matching Problems

Multi-Agent Systems (MAS) have been applied in recent years to assignment or matching problems in order to enhance privacy in preferences and constraints for individuals, and to facilitate the distribution of solving. A further step in this direction consists in using the organisational structures provided by MAS. Thus, in this paper, we rely upon the capability of Multilevel MAS to reify intermediate viewpoints between the individual and the collective levels, in order to encompass matching or assignment problems. Therefore we define a meta-model for assessing the welfare of agent groups with respect to relevant metrics, so that these groups are able to elaborate solutions that improve the collective well-being without forcing them to disclose all their private information. Finally, we outline the general principles for distributed solvers designed for this type of modelling.