Mechanism Design for Preference Aggregation over Coalitions

The aggregation of individuals’ preferences into a single group outcome is both well-studied and fundamental within decision theory. Historically, though, a pervasive simplification has been to strip agents of the ability to form coalitions and strategically reveal their intentions. CP techniques can address such possibilities within a restricted socialchoice framework that represents mechanisms for aggregating preferences as decision trees; the internal nodes are labeled by various individuals, and the leaves represent outcomes that arise from the flow of individual decisions down the various branches of the tree. Currently, a highly simplified version of the language allows the automatic generation of such trees meeting user-specified properties, using a standard CSP-solver. By further enrichments within a more specialized reasoning framework, the eventual goal is to circumvent negative theoretical theorems concerning preference aggregation, by exploiting the combination of coa...

• Type: Conference
• Year: 2005
• Where: CP
• Authors: Eric I. Hsu, Sheila A. McIlraith
• Keywords: CP 2005, Negative Theoretical Theorems, Programming Languages, Single Group Outcome, Specialized Reasoning Framework