Evaluating Information Variation in Informed Agent Negotiation

Much of the B2B real world commercial activities today require the resolution of multiple issues before a trade is agreed and contract signed. Negotiation is typically the way in which these multi-issue transactions are resolved. For B2B e-commerce, the automation of this negotiation for trade has the potential to revolutionise the way that business is conducted. Intelligent agents have been the focus for developments into automating e-commerce trade negotiation, however, much research is still required before automated negotiation is possible. E-commerce negotiation occurs on the Internet, a dynamic, information rich domain. Information in negotiation is crucial, and a negotiating agent should account for changes in information in its strategy. An intelligent agent framework for negotiation has been developed based on a maximal entropy approach which centers decision making on the certainty of information. We evaluate how the variations in information and certainty affect an agent negotiating in an exemplar negotiation scenario.