Covering the Unseen: Information Demand Coverage Optimization for Retrieval-Augmented Generation

2026-06-28

AI Summary

This work addresses a key limitation of traditional retrieval-augmented generation (RAG) systems, which often overlook critical sub-questions in complex queries due to reliance on a single query embedding. The authors formalize context selection as an information-need coverage optimization problem, proposing an unsupervised, training-free strategy that first generates diverse sub-queries and then constructs a multi-dimensional information-need distribution via inverse validation weighting. Optimal context sets are selected by minimizing the Sinkhorn-Wasserstein distance to this distribution. This approach is the first to model information needs as a multi-dimensional distribution and offers theoretical guarantees while remaining agnostic to the underlying retriever. Evaluated on six open-domain question answering benchmarks, it achieves substantial gains of 6.5–9.7 points in Exact Match over standard top-k retrieval and significantly outperforms strong baselines including MMR, DPP, and BGE-Reranker.

Abstract

Retrieval-augmented generation (RAG) typically treats context selection as ranking chunks against a single query embedding. This assumption breaks down for complex queries, such as multi-hop or ambiguous questions, where top-k selection tends to over-cover one semantic aspect while ignoring critical sub-questions. We propose GeoRAG, which recasts context selection as Information Demand Coverage Optimization. GeoRAG builds a multi-dimensional demand distribution through diverse sub-query generation and reverse-validation weighting, then selects context by minimizing the Sinkhorn-Wasserstein distance between this demand distribution and the coverage of the selected set. The resulting demand-weighted facility-location objective is monotone submodular, giving a $1-1/e$ greedy guarantee, which we approximate with a Sinkhorn-based marginal-gain surrogate. The method is unsupervised, training-free, and retrieval-agnostic. We further show that single-point, query-proximity scorers cannot cover multi-modal demands, exposing a structural limit of ranking-based selection. On six open-domain QA benchmarks, GeoRAG improves exact match (EM) by +6.5 to +7.5 points over top-k truncation (up to +9.7 on HotpotQA and ASQA) and outperforms strong baselines including MMR, DPP, BGE-Reranker, SMART-RAG, and AdaGReS, with stable gains across context budgets and sub-query generators.

Problem

• Retrieval-Augmented Generation
• Information Demand Coverage
• Multi-hop Questions
• Context Selection
• Ambiguous Queries

Innovation

• Information Demand Coverage
• Sinkhorn-Wasserstein Distance
• Submodular Optimization
• Retrieval-Augmented Generation
• Multi-hop Question Answering