Generalizing Preference Elicitation in Combinatorial Auctions

Combinatorial auctions where agents can bid on bundles of items are desirable because they allow the agents to express complementarity and substitutability between the items. However, expressing one’s preferences can require bidding on all bundles. Selective incremental preference elicitation by the auctioneer was recently proposed to address this problem but the idea was not evaluated. In this paper we show that automated elicitation is extremely beneficial: as the number of items for sale increases, the amount of information elicited is a small and diminishing fraction of the information collected in traditional “direct revelation mechanisms” where bidders reveal all their valuation information. The elicitors also maintain the benefit as the number of agents increases—except rank lattice based elicitors which we show ineffective. We also develop elicitors that combine different query types, and we present a new query type that takes the incremental nature of elicitation to a new level by allowing agents to give approximate answers that are refined only on an as-needed basis. We show that determining VCG payments requires very little additional elicitation. Finally, we show that elicitation can be easily adapted to combinatorial reverse auctions, where the benefits are similar to those in auctions, except that the elicitation ratio improves as the number of agents increases. In the process, we present methods for evaluating different types of elicitation policies.