A Meta-strategy for Multi-issue Negotiation

Abstract

Automated negotiation is a process where multiple software agents interact with each other over certain matters through making offers and counteroffers. This paper considers a buyer agent negotiating with a seller agent over multiple issues. A meta-negotiation strategy is proposed to improve the negotiation process outcomes. The presented negotiation strategy is a cooperative negotiation one that uses both, preference-based and fuzzy similarity mechanisms. The preference-based mechanism is used for generating quantitative offers while the fuzzy similarity is used for in the process of generating qualitative offers. The preference-based mechanism considers the preferences of the opponent when making offers; the agent makes more concessions on the negotiation issues that are believed to be more preferred to the opponent. In the fuzzy similarity approach, the agent formulates an offer that is believed to be most similar to the one received from the opponent in the last negotiation round. The experimental results show that the proposed cooperative negotiation strategy outperforms the basic negotiation strategy in all the evaluation criteria such as utility, agreement rate and Nash product.