The aggregate power of all resources on the Internet is enormous. The Internet can be viewed as a massive virtual organization that holds tremendous amounts of information and resources with different ownerships. However, little is known about how to run this organization efficiently. This dissertation studies the problems of distributed task and resource allocation among self-interested agents in virtual organizations. The developed solutions are not allocation mechanisms that can be imposed by a centralized designer, but decentralized interaction mechanisms that provide incentives to self-interested agents to behave cooperatively. These mechanisms also take computational tractability into consideration due to the inherent complexity of distributed task and resource allocation problems. Targeted allocation mechanisms can achieve global task allocation efficiency in a virtual organization and establish stable resource-sharing communities based on agents’ own decisions about whether or not to behave cooperatively. This high level goal requires solving the following problems: synthetic task allocation, decentralized coalition formation and automated multiparty negotiation. For synthetic task allocation, in which each task needs to be accomplished by a virtual team composed of self-interested agents from different real organizations, my approach is to formalize the synthetic task allocation problem as an algorithmic mechanism design optimization problem. I have developed two approximation mechanisms that I prove are incentive compatible for a synthetic task allocation problem. This dissertation also develops a decentralized coalition formation mechanism, which is based on explicit negotiation among self-interested agents. Each agent makes its own decisions about whether or not to join a candidate coalition. The resulting coalitions are stable in the core in terms of coalition rationality. I have applied this mechanism to form resource sharing coalitions in computational grids and buyer coalitions in electronic markets. The developed negotiation mechanism in the decentralized coalition formation mechanism realizes automated multilateral negotiation among self-interested agents who have symmetric authority (i.e., no mediator exists and agents are peers). In combination, the decentralized allocation mechanisms presented in this dissertation lay a foundation for realizing automated resource management in open and scalable virtual organizations.

Le travail que nous présentons dans cette thèse s'articule autour du problème de la formation de coalitions entre des agents égoïstes qui planifient leurs activités, dans les systèmes multi-agents (SMA). Nous avons proposé, dans un premier temps, un mécanisme qui se fonde sur l’analyse des actions des agents dans leurs plans et le raisonnement sur les plans des autres, grâce notamment au calcul d’un degré de croyance sur les actions. Nous nous sommes, par ailleurs, intéressés au problème de la formation de coalitions avec des contraintes dynamiques et des préférences que les agents révèlent et communiquent aux autres lors de leurs négociations. Enfin, nous avons affiné notre mécanisme de formation des coalitions en permettant une recherche des coalitions guidée par la construction d'un arbre de contraintes et d'un arbre de coalitions, qui sont ensuite exploré par le biais de l'algorithme Monte-Carlo. / The work we present, in this thesis, focuses on the coalition formation problem for self-interested agents which plan their activities in multi-agents systems. As a first step, we have proposed, a mechanism that is based on the analysis of the agents' actions in their plans and reasoning about the plans of others. Additionally, we have addressed the problem of coalition formation with dynamic constraints and preferences that agents reveal and communicate to others during their negotiations. Finally, we have refined our coalition formation mechanism allowing a guided search of the coalitions by building a tree of constraints and a tree of coalitions. Each tree is explored by means of the Monte-Carlo algorithm.