Arbitration and Matchmaking for Agents with Conflicting Interests

The exchange of goods and services among software agents requires reliable and fair brokering mechanisms to match trading parties and to mediate among their conflicting interests. Available trading models for electronic marketplaces are fixed price selling, bilateral multi-step negotiations, and various forms of auctioning. These models demand trading parties to evaluate appropriate interest matches on their own and encourage them to pretend inexact interests to their advantage. We introduce an arbiter as intermediary that finds buyers and suppliers with best matching interests. The intermediary uses matching and arbitration protocols that ensure better overall benefit than random matches, avoid advantages for agents that manipulate their interests (lies), preserve the mutual privacy of interests of the trading parties, and, if desired, their anonymity. We analyse the protocols with respect to their applicability under various conditions, investigate their robustness with different utility distributions by simulations, and describe which forms of interest manipulations can be avoided.