Intelligent Agents for Automated One-to-Many E-Commerce Negotiation

Negotiation is a process in which two or more parties with different criteria, constraints, and preferences, jointly reach an agreement on the terms of a transaction. Many current automated negotiation systems support one-to-one negotiation. One-to-many negotiation has been mostly automated using various kinds of auction mechanisms, which have a number of limitations such as the lack of the ability to perform two-way communication of offers and counteroffers. Moreover, in auctions, there is no way of exercising different negotiation strategies with different opponents. Even though auction-based online trading is suitable for many applications, there are some in which there is a need for such greater flexibility. There has been a significant body of work towards sophisticated one-to- one automated negotiation. In this paper, we present a framework for one-to-many negotiation by means of conducting a number of concurrent coordinated one-to-one negotiations. In our framework, a number of agents, all working on behalf of one party, negotiate individually with other parties. After each negotiation cycle, these agents report back to a coordinating agent that evaluates how well each agent has done, and issues new instructions accordingly. Each individual agent conducts reasoning by using constraint-based techniques. We outline two levels of strategies that can be exercised on two levels, the individual negotiation level, and the coordination level. We also show that our one-to-many negotiation architecture can be directly used to support many-to-many negotiations. In our prototype Intelligent Trading Agency (ITA), agents autonomously negotiate multi- attribute terms of transactions in an e-commerce environment tested with a personal computer trading scenario