The paper presents a prototype of Fuzzy e-Negotiation Agents (FeNAs) for autonomous multi-issue negotiation in e-commerce. It considers negotiation as a form of distributed decision making in the presence of limited common knowledge and imprecise/soft constraints that can be modeled as a distributed fuzzy constraint satisfaction problem (DFCSP). FeNAs incorporate the principles of utility theory within DFCSPs and use fuzzy constraint-based reasoning in order to find a consensus that maximizes the agent's utility and the level of its fuzzy constraint satisfaction subject to its acceptability by other agents. The paper presents aspects of problem representation and negotiation mechanisms used by FeNAs in the context of DFCSPs. An overview of FeNAs is provided and some capabilities for automated multi-issue negotiation are illustrated with two scenarios of e-commerce trading.Keywords e-negotiation agents, Fuzzy constraint-based reasoning
IntroductionNegotiation is a common form of decision-making where two or more parties jointly explore possible agreements in order to reach a consensus [37]. Negotiation has traditionally been a subject of study in the game-theoretic [37, 38], economics [14] and management science [29] research. It has also been an active area of research in Artificial Intelligence (AI) and in particular in distributed AI (DAI) and multi-agent systems (MAS) (e.g. [28,34,42,43,44,47]). The increased potential of AI technology in supporting and automating negotiation has been recognized in a wide range of real-world problems including group conflict resolution [33,44], business negotiations [13], resource allocation and scheduling [39], and recently in e-commerce [4,15, 22, 38].E-commerce offers new channels and business models for buyers and sellers to effectively and efficiently trade goods and services in on-line markets. It has now embraced so called dynamic e-commerce, which enables buying and selling goods and services through flexible pricing mechanisms that change with supply and demand. Presently, dynamic e-commerce manifests itself most commonly in exchanges and auctions that focus on basic price and quantity bidding. As the market quickly evolves, new advanced dynamic e-commerce solutions emerge (called also negotiated e-commerce or e-negotiation) to enable mapping complex negotiation models in business transactions to e-commerce. These models can include direct negotiations, multi-stage negotiations, tendering and multiparameter bidding, collaborative trading and aggregations.As e-commerce becomes common a large number of Internet-based systems have been developed to assist users in different aspects of electronic trading. In particular systems that use software agent technologies are proving to be effective in supporting and automating various stages of the trading process (e.g. product finding, supplier finding, product ordering, delivery monitoring) [3,15, 38]. It has also been recognized that software agents can play an important role in providing automation support for the negotiatio...