Building Relationships and Negotiating Agreements in a Network of Agents

Abstract

Fully automated trading, such as e-procurement, is virtually unheard of today. Trading involves the maintenance of effective business relationships, and is the complete process of: need identification, product brokering, supplier brokering, offer-exchange, contract negotiation, and contract execution. Three core technologies are needed to fully automate the trading process. First, real-time data mining technology to tap information flows and to deliver timely information at the right granularity. Second, intelligent agents that are designed to operate in tandem with the real-time information flows from the data mining systems. Third, virtual institutions in which informed trading agents can trade securely both with each other and with human agents in a natural way. This discussion focusses on the second technology - the design of “information driven” trading agents. The “information-based agency” is being developed principally by John Debenham [UTS, Australia] and Carles Sierra [IIIA, Spain]. Published work to date has considered argumentation frameworks between a set of (often just two) agents, and has not taken into account the relationships that the various agents have with other agents. Drawing ideas from the business research literature we are constructing a model of multi-agent interaction that accommodates the sorts of inter-agent agreements found in the business world. The working hypothesis is that the information-based approach will prove to be a suitable conceptual framework through which networks of agents, interacting under the constraints of various styles of business relationship, may be implemented.