Documents
Resources
Learning Center
Upload
Plans & pricing Sign in
Sign Out
Your Federal Quarterly Tax Payments are due April 15th Get Help Now >>

Preference Handling in Combinatorial Domains: From AI to Social Choice

VIEWS: 13 PAGES: 11

In both individual and collective decision making, the space of alternatives from which the agent (or the group of agents) has to choose often has a combinatorial (or multiattribute) structure. We give an introduction to preference handling in combinatorial domains in the context of collective decision making and show that the considerable body of work on preference representation and elicitation that AI researchers have been working on for several years is particularly relevant. After giving an overview of languages for compact representation of preferences, we discuss problems in voting in combinatorial domains and then focus on multiagent resource allocation and fair division. These issues belong to a larger field, which is known as computational social choice and which brings together ideas from AI and social choice theory, to investigate mechanisms for collective decision making from a computational point of view. We conclude by briefly describing some of the other research topics studied in computational social choice. [PUBLICATION ABSTRACT]

More Info
									                                                                                                                                   Articles




                                  Preference Handling
                              in Combinatorial Domains:
                                From AI to Social Choice

                                                     Yann Chevaleyre, Ulle Endriss,
                                                   Jérôme Lang, and Nicolas Maudet




                                                     I
       n In both individual and collective
       decision making, the space of alterna-
       tives from which the agent (or the group          ndividual decision making is mostly guided by the agent’s preferences
       of agents) has to choose often has a          over his or her possible decisions. Similarly, group decision making is
       combinatorial (or multiattribute) struc-      guided by the preferences of the agents in the group. For instance, when
       ture. We give an introduction to prefer-      a group of autonomous agents need to agree on an allocation of resources
       ence handling in combinatorial do-
                                                     among themselves, then each individual will judge the outcome accord-
       mains in the context of collective
       decision making and show that the             ing to his or her own preferences and will have to transmit parts of these
       considerable body of work on preference       preferences (possibly indirectly and possibly reluctantly so) to his or her
       representation and elicitation that AI        peers in the process of negotiation. Also, to be able to assess whether the
       researchers have been working on for          negotiation outcome should be considered a “good” allocation (say,
       several years is particularly relevant.       whether it reflects a fair agreement) requires knowledge of the individual
       After giving an overview of languages         preferences. Similarly, when voting on a proposition or for a candidate,
       for compact representation of prefer-         the ballot submitted by each individual reflects some aspect of his or her
       ences, we discuss problems in voting in       own preferences, and the voting protocol in place is charged with aggre-
       combinatorial domains and then focus
                                                     gating these preferences into a decision that (we hope) constitutes a good
       on multiagent resource allocation and
       fair division. These issues belong to a
                                                     reflection of the collective will of the population.
       larger field, which is known as compu-           The classical discipline concerned with the study of mechanisms for
       tational social choice and which brings       collective decision making is social choice theory (Arrow, Sen, and Suzu-
       together ideas from AI and social choice      mura 2002). Much work in the field has concentrated on normative ques-
       theory, to investigate mechanisms for         tions and on establishing abstract results regarding the possibility of
       collective decision making from a com-        designing mechanisms meeting certain requirements. For instance, a
       putational point of view. We conclude         seminal result in the field, Arrow’s Impossibility Theorem, shows that
       by briefly describing some of the other       there can exist no preference-aggregation mechanism that would simul-
       research topics studied in computation-
                                                     taneously satisfy a small number of natural requirements (for example,
       al social choice.
                                                     the aggregation function shouldn’t be dictatorial). Computational con-




Copyright © 2008, Association for the Advancement of Artificial Intelligence. All rights reserved. ISSN 0738-4602     WINTER 2008 37
Articles

               cerns, however, have mostly been neglected: What        clude by mentioning some of the other topics that
               is the computational complexity of the mecha-           have recently been addressed in the computation-
               nisms proposed by social choice theorists? What         al social choice literature.
               are the appropriate algorithmic techniques for
               these problems? What happens if the number of
               alternatives to choose from becomes very large?
                                                                                  Preferences in
                  Such considerations have given rise to an inter-             Combinatorial Domains
               disciplinary research effort at the interface of AI
                                                                       Collective decision making in combinatorial
               and computer science with social choice theory,
                                                                       domains first and foremost requires modeling the
               sometimes dubbed computational social choice. On
                                                                       preferences of individual decision makers over
               the one hand, computational social choice is con-
                                                                       alternatives with a combinatorial structure. In our
               cerned with the application of techniques devel-
                                                                       discussion of preference-representation languages,
               oped in computer science, such as complexity
                                                                       we start by listing some natural requirements for
               analysis or algorithm design, to the study of social
                                              
								
To top