DMCA
Partial-revelation VCG mechanism for combinatorial auctions
| Venue: | In Proceddings of the National Conference on Artificial Intelligence (AAAI |
| Citations: | 43 - 18 self |
BibTeX
@INPROCEEDINGS{Conen_partial-revelationvcg,
author = {Wolfram Conen and Tuomas Sandholm},
title = {Partial-revelation VCG mechanism for combinatorial auctions},
booktitle = {In Proceddings of the National Conference on Artificial Intelligence (AAAI},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Winner determination in combinatorial auctions has received significant interest in the AI community in the last 3 years. Another difficult problem in combinatorial auctions is that of eliciting the bidders ’ preferences. We introduce a progressive, partial-revelation mechanism that determines an efficient allocation and the Vickrey payments. The mechanism is based on a family of algorithms that explore the natural lattice structure of the bidders ’ combined preferences. The mechanism elicits utilities in a natural sequence, and aims at keeping the amount of elicited information and the effort to compute the information minimal. We present analytical results on the amount of elicitation. We show that no valuequerying algorithm that is constrained to querying feasible bundles can save more elicitation than one of our algorithms. We also show that one of our algorithms can determine the Vickrey payments as a costless by-product of determining an optimal allocation.
Keyphrases
combinatorial auction partial-revelation vcg mechanism vickrey payment winner determination natural sequence feasible bundle significant interest bidder preference efficient allocation natural lattice structure present analytical result difficult problem information minimal costless by-product ai community partial-revelation mechanism optimal allocation
