Using Recommendation to Improve Negotiations in Agent-based Systems

Mateusz Lenar (Wroclaw University of Technology)

Janusz Sobecki (Wroclaw University of Technology, Poland)

Abstract: In this paper we present research works on non-intuitive and low-efficient negotiations between agents in agent based system. We find recommendation techniques as a suitable method of making negotiation smarter and more efficient. We have introduced into the negotiation thread the improvements in the form of hybrid recommendation. Hybrid recommendation is composed of three basic elements: demographic, collaborative and content-based. On the base of a negotiation algorithm we have studied how recommendation methods can improve the whole process of finding mutually acceptable agreements between agents. The proposed methodology is presented using a travel agency and its client negotiation.

Keywords: Petri Nets, agent-based systems, hybrid recommendation, negotiation thread