A two-step approach to building bilateral consensus between agents based on relationship learning theory

Researchers are increasingly focusing on the agent based approach to transaction support in ubiquitous commerce. These agents work autonomously to maximize utility on the user’s behalf. In the case of a cooperative game, rather than a win–lose zero-sum game, agents may negotiate with each other or have a negotiating agent provide a suggestion that can be reasonably accepted by the dyad to build a consensus. In this paper we propose a novel methodology that increases agent performance in terms of costs associated with building consensus and successful negotiation rates. To do so, we develop a two-step approach: joint learning and negotiation to consensus building. We also conduct an experimental study to show the feasibility of the methodology.