Preference Elicitation and Allocation with Read-Once Preferences [2 citations — 0 self]
|
Abstract:
Elicitation of real-valued preferences is a key capability in many allocation problems, especially in electronic commerce. When multiple items are to be allocated, preference elicitation is difficult because the parties ' preferences over items are generally not additive due to complementarityand substitutability. In this paper we study the elicitation of real-valued preferences over subsets of items, showing that read-once preferences (a natural restriction in many applications that is still powerful enough to capture complementarity and substitutability) can be elicited in a polynomial number of value queries. Such elicitation can be used by a shopping agent to elicit its user's preferences. It can also be used to elicit bidders ' preferences in a combinatorial auction. We also show that if a party's preferences are close to read-once, then a good approximation of the preferences can be elicited quickly. Finally, we study the complexity of the computational problem of allocating the items given the parties ' preferences. We show that this is NP-hard even with just two parties with read-once valuations. However, in the natural setting where only one of the parties has a read-once valuation and the other is additive (for example, a shopping agent purchasing items individually and then bundling them to give the user), the allocation problem is solvable in polynomial time. 1
Citations
| 528 | Queries and concept learning – Angluin - 1988 |
| 131 | Iterative Combinatorial Auctions: Theory and Practice – Parkes - 2001 |
| 102 | Learning read-once formulas with queries – Angluin, Hellerstein, et al. - 1993 |
| 69 | Preference elicitation in combinatorial auctions – Sandholm, Boutilier - 2006 |
| 46 | Sandholm: “Partial-revelation VCG Mechanism for Combinatorial Auctions – Conen, T - 2002 |
| 36 | Linear programming and Vickrey auctions – Bikhchandani, Vries, et al. - 2001 |
| 31 | The communication complexity of efficient allocation problems – Nisan, Segal - 2002 |
| 28 | Learning arithmetic read-once formulas – Bshouty, Hancock, et al. - 1995 |
| 23 | Effectiveness of preference elicitation in combinatorial auctions – Hudson, Sandholm - 2002 |
| 9 | An algorithm to learn read-once threshold formulas, and transformations between learning models – Bshouty, Hancock, et al. - 1994 |
