Elicitating, modeling, and processing uncertain human preferences for software agents in electronic negotiations: An empirical study

Transaction costs in electronic business may become marginal, since negotiation processes can be assigned to software agents which act autonomously on behalf of their human principals. This advantage makes electronic negotiations highly appealing to actors in various e-business areas, including online auctions and product configuration negotiations. However, software agents can negotiate appropriately only if the preferences of their principals are explicitly available. While this task has been acknowledged as crucial challenge, it has only rarely been investigated how preferences of human principals can be extracted and modeled. This paper addresses this research gap by suggesting, implementing, and empirically testing a preference elicitation method that is based on fuzzy set theory and accounts for the impreciseness and subjectivity of preferences. Our findings indicate that the proposed method is suited for elicitating fuzzy set based preferences. Furthermore, the fuzzy preference model can project human decisions more accurately than a traditional, crisp approach.