Learning Marginal Viewpoint Utility for Budgeted Selection in RAG

Retrieval-augmented generation over multi-viewpoint queries must preserve distinct answers and positions from the retrieved evidence while operating under a fixed context budget. We introduce LMVU, a lightweight ranker that learns to prioritize candidate passages by the additional viewpoint coverage they contribute to the selected context. Training uses viewpoint annotations on training examples from each benchmark to compare candidate passages added to the same context. At inference, LMVU repeatedly scores passages that fit the remaining context budget as the context grows, without accessing these annotations. We evaluate recovered viewpoint coverage in terms of breadth (${\mathrm{RVC}}_1$) and repeated support from distinct sources (${\mathrm{RVC}}_2$). Exact oracles measure the maximum coverage achievable from the same evidence pool within the same budget. Across four viewpoint-annotated benchmarks at a 1,024-token context budget, LMVU achieves the highest average coverage on both metrics. It exceeds the strongest non-oracle baseline average by $0.037$ ${\mathrm{RVC}}_1$ points and $0.033$ ${\mathrm{RVC}}_2$ points, and reaches the exact viewpoint-breadth ceiling on two benchmarks.