Accelerated Preference Elicitation with LLM-Based Proxies

2025-01-24

AI Summary

In combinatorial auctions, bidders face difficulties in expressing complex preferences, while conventional preference elicitation methods suffer from excessive query rounds, high cognitive load, and low efficiency. Method: This paper proposes a natural-language interactive preference elicitation framework that jointly leverages large language models (LLMs) and disjunctive normal form (DNF) modeling. It integrates LLMs’ semantic understanding with DNF’s formal expressiveness to construct a lightweight, interpretable, and verifiable proxy architecture. A natural-language sandbox interface and query optimization strategy are designed to minimize interaction complexity without compromising theoretical soundness. Contribution/Results: Experiments demonstrate that the method achieves near-optimal resource allocation using only ~20% of the queries required by classical adaptive elicitation approaches—significantly reducing bidder burden and improving mechanism efficiency.

Technology Category

• Multiagent Systems: Mechanism Design
• Machine Learning: Learning Preferences or Rankings
• Game Theory and Economic Paradigms: Mechanism Design

Application Category

• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Bidders in combinatorial auctions face significant challenges when describing their preferences to an auctioneer. Classical work on preference elicitation focuses on query-based techniques inspired from proper learning--often via proxies that interface between bidders and an auction mechanism--to incrementally learn bidder preferences as needed to compute efficient allocations. Although such elicitation mechanisms enjoy theoretical query efficiency, the amount of communication required may still be too cognitively taxing in practice. We propose a family of efficient LLM-based proxy designs for eliciting preferences from bidders using natural language. Our proposed mechanism combines LLM pipelines and DNF-proper-learning techniques to quickly approximate preferences when communication is limited. To validate our approach, we create a testing sandbox for elicitation mechanisms that communicate in natural language. In our experiments, our most promising LLM proxy design reaches approximately efficient outcomes with five times fewer queries than classical proper learning based elicitation mechanisms.

Problem

• Combinatorial Auctions
• Preference Elicitation
• Efficient Allocation

Innovation

• LLM-based Intelligent Assistant
• Combination Auctions
• Reduced Query Quantity