An Agent-based Multi-attribute Sealed-bid Design for Bilateral Contract

This paper presents an agent-based sealed-bid design for multi-issue negotiation in an incomplete information setting. The information of both agents is considered completely private. In our agent-based model, each agent builds a multi-dimensional fuzzy satisfaction set for the attributes of a commodity using fuzzy linguistic terms. The issues relating to an offer are divided into two categories (i.e., quantitative issues and qualitative issues). At each time-step of the negotiation, both agents submit a combined offer of quantitative issues for each discrete value of qualitative issue to the mediate agent simultaneously. This method offers flexibility for the agents represent their satisfaction for the combinations of different attributes using a multi-dimensional membership functions, taking into account the interdependencies between the attributes. In addition, using the proposed method of calculating agreed-price, agents can be stimulated to reach an agreement as early as possible. Therefore, this method may shorten negotiation process, which is desired for both parties. And the design can discourage counter-speculation and effectively control fraud and misrepresentation in a certain extent. Through a case study, the capabilities of the proposed method are illustrated.