Dependence in games and dependence games
Autonomous Agents and Multi-Agent Systems
Argumentation theoretic foundations for abstract dependence networks
AT'13 Proceedings of the Second international conference on Agreement Technologies
| Hi-index | 0.00 |
It is desirable that artificial agents can help each other when they cannot achieve their goals, or when they profit from social exchanges. In this work we study coalition formation processes supported by enforced agreements and we define two qualitative criteria, the do-ut-des property and the composition property, that establish when a coalition is admissible to be formed. The do-ut-des property is based on a balance between the advantages and the burdens of an agent, when it agrees an enforced agreements. The composition property is a refinement of the do-ut-des property that takes into account also the costs and the risks deriving from the coalition formation process.Two relevant aspects distinguish our approach from the solution criteria developed in cooperative game theory. First, the do-ut-des property and the composition property are not based on an explicit utility function associated to the goals of an agent, and hance they can be used also in that cases in which the importance that agents give to their own goals is unknown. Second, a coalition has all the necessary information to establish if it satisfies the do-ut-des property or the composition property, therefore these two properties can be used in the case not all the space of possible coalitions is known.