Strategies for offer generation and relaxation in fuzzy constraint-based negotiation models

Issue title: Advances in Agent-mediated Automated Negotiations

Guest editors: Minjie Zhangx and Takayuki Itoy

Article type: Research Article

Authors: Lopez-Carmona, Miguel A.^{; *} | Marsa-Maestre, Ivan | Velasco, Juan R. | de la Hoz, Enrique | Alarcos, Bernardo

Affiliations: Department of Computer Engineering, Escuela Politecnica Superior, Universidad de Alcala, Alcala de Henares (Madrid), 28871, Spain | [x] University of Wollongong, Wollongong, NSW 2522, Australia | [y] Nagoya Institute of Technology, Gokiso, Showa, Nagoya 466-8555, Japan

Correspondence: [*] Corresponding author. Tel.: +34 91 8856673; Fax: +34 91 8856641; E-mail: miguelangel.lopez@uah.es

Abstract: Fuzzy constraints have been used in several automated multi-attribute negotiation models. It is recognized that fuzzy constraints represent an efficient way of capturing requirements and preferences, and an useful mechanism for representing trade-offs. Most approaches are mainly focused on using constraints as a framework for describing preferences, and few works focus on using them as an element in the communication process itself in order to elaborate efficient negotiation strategies. This paper proposes and evaluates a set of strategies for offer generation and for the construction of offer relaxation requests in constraint-based negotiation models. To deploy the strategies, a fuzzy constraint based model for non-mediated bilateral automated purchase negotiations has been used. Fuzzy constraints are used both to represent preferences and to express offers. A set of locutions and decision mechanisms which trigger them are fully specified, where each agent may decide its degree of cooperation and its degree of expressiveness. Expressiveness is based on the propagation of constraints and relaxation requests. The paper analyzes which combination of different agents' attitudes allow to improve the negotiation processes. Experimental evaluation confirms the advantage of an expressive approach based on the propagation of constraints. In addition, this paper studies how applying a clustering algorithm to the seller's catalogue of products instead of managing single products may contribute to an improvement in the duration of the negotiation dialogues, and to a significant improvement on the utility of the deals that are achieved. These results support the usefulness of the proposed negotiation strategies.

Keywords: Multi-attribute negotiation, automated purchase negotiation, fuzzy constraints, multi-agent systems

DOI: 10.3233/MGS-2010-0163

Journal: Multiagent and Grid Systems , vol. 6, no. 5-6, pp. 503-525, 2010