Preference Elicitation in Combinatorial Auctions (Extended Abstract)

摘要: Combinatorial auctions (CAs) where bidders can bid on bundles of items be very desirable market mechanisms when the sold exhibit complementarity and/or substitutability, so bidder’s valuations for are not additive. However, in a basic CA, may need to exponentially many bundles, leading diculties determining those valuations, undesirable information revelation, and unnecessary communication. In this paper we present design an auctioneer agent that uses topological structure inherent problem reduce amount it needs from bidders. An analysis tool is presented as well data structures storing optimally assimilating received Using information, then narrows down set (welfare-maximizing or Pareto-ecient) allocations, decides which questions ask next. Several algorithms value, order, rank information. A method making elicitor incentive compatible.