Using negotiation and coordination in multiagent Intelligent Cooperative Information Systems

Abstract

Users of future Intelligent Cooperative Information Systems (ICIS) will be able to have their queries answered from a variety of diverse database and knowledge-based sources. In order to allow these existing knowledge-based software systems to work together in a cooperative fashion, several key techniques from the field of Distributed Artificial Intelligence (DAI) will be required. Specifically, this paper looks at negotiation as a means of coordinating “agents” in multiagent cooperative problem solving (CPS) systems. A novel form of incremental negotiation called knowledge-based negotiation is presented. Agent conflicts are resolved through a negotiation process which makes use of shared agent knowledge representations called shareable perspectives. The combination of two shareable agent perspectives forms an interagent issue relation which relates the agents to a single domain object by means of agent relevant domain aspects. The pairings of interagent issue relations are grouped into a relational network which is maintained by a third-party arbitrator agent. The arbitrator uses this relational network to develop alternative suggestions for the conflicting agents during mediation. The arbitrator reviews the negotiation dialog for relevant issues between the agents and then searches the network for relations that are known to exist between the agents. The arbitrator presents these relations to the agents who use them to identify and rank other possible viable alternatives which they might not have considered earlier. If agents still do not agree (mediation fails), binding arbitration techniques are employed. Both arbitrator agent mediation and arbitration algorithms are described with examples. The resulting research is implemented as the Designer Fabricator Interpreter (DFI) System which integrates diverse agent knowledge of the construction domain into a “what-if ” tool used to critiquing connection designs in buildings at preliminary design time.KeywordsNegotiation ProcessDomain ObjectConflicting AgentAgent IssueAgent PerspectiveThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.