ICMAS '96: Norms, Obligations, and Conventions

tant will behave as a shark with regard to potential partners, always looking for the most convenient transaction, thereby infringing on existing commitments. Other difficult tasks, more generally, are how to obtain a robust performance in teamworks (Cohen and Levesque 1990); how to prevent agents from dropping their commitments ; or better, how to regulate agents dropping their commitments to a joint action to not disrupt the common activity and preclude the common goal being achieved (Jen-These tasks have now entered the MAS field's common knowledge. Other problems are perhaps less obvious. For example, the existence of so-called T he use of a normative vocabulary and concepts (obligations, deontic operators, norms, and so on), as well as the insertion of norms in knowledge bases for expert systems, has a well-established tradition in many AI subfields (legal expert systems, artificial normative reasoning , and so on). In adjacent domains (logical philosophy, social philosophy, decision theory), both legal and social norms have received considerable, if not satisfactory, attention. The advent of large communication networks, civic networks, and so on, has contributed dramatically to the attention given by the scientific community to several normative issues: authorization, access regulation, privacy maintenance, respect of decency , and so on, not to mention the more obvious problems associated with the regulation of the use and purposes of networks. In the multiagent system (MAS) discipline, social norms and laws are perceived to help improve coordina-efforts done by MAS researchers and designers to construct autonomous agents (Wooldridge and Jennings 1995) carry a number of interesting but difficult tasks, including how to avoid interferences and collisions among agents moving around and separately acting in a common space and how to ensure that negotiations and transactions fulfill the norm of reciprocity. Imagine a software assistant delegated to conduct transactions on behalf of its user. Because of its loyalty (benevolence), the assis-model of authorization and (institutional) empowerment (Jones and Ser-got 1995). That social norms are of vital importance for AI is witnessed by the variety of subcommunities dealing with normative concepts and questions: legal expert systems, artificial normative reasoning, AI and law, deontic logics for computer science, MAS, and distributed artificial intelligence (DAI) are some examples. However, these communities suffer from a poor level of communication and confrontation among one another. Bringing together people from these communities was one major reason of interest behind the Second International Conference on Multiagent Systems …