BEAR: Budgeted Evidence Allocation for Multi-Document Reasoning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sun, Lin, Zhang, Linglin, Huang, Jingang, Jia, Change, Cheng, Zhengwei, Zhang, Xiangzheng
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917537139130368
author Sun, Lin
Zhang, Linglin
Huang, Jingang
Jia, Change
Cheng, Zhengwei
Zhang, Xiangzheng
author_facet Sun, Lin
Zhang, Linglin
Huang, Jingang
Jia, Change
Cheng, Zhengwei
Zhang, Xiangzheng
contents We argue that multi-document reasoning is constrained not only by how much text a model can read, but also by how limited query-time evidence budget is allocated across documents and semantic granularities. Full-context inference exposes the model to broad evidence non-selectively and at high per-query cost, while flat chunk retrieval often returns locally relevant passages that are weakly organized for cross-document synthesis. We present \textbf{BEAR}, a framework for structured evidence allocation that builds hierarchical semantic indices offline and performs coarse-to-fine evidence access at query time through complementary \emph{exploration} and \emph{recovery} paths. This coarse-to-fine design can be viewed as structured evidence allocation under a fixed evidence-context budget. Across synthetic and real-world benchmarks, BEAR performs particularly strongly on DragonBall, remains competitive with strong retrieval-based baselines on HotpotQA, and yields the best retrieval-based result on 2Wiki under our evaluated protocol, while operating under substantially smaller \emph{query-time evidence budgets} than the reported long-context references. Additional analyses suggest that the gains are associated with hierarchy as an allocation substrate together with complementary exploration and recovery, rather than semantic chunking alone.
format Preprint
id arxiv_https___arxiv_org_abs_2601_18116
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle BEAR: Budgeted Evidence Allocation for Multi-Document Reasoning
Sun, Lin
Zhang, Linglin
Huang, Jingang
Jia, Change
Cheng, Zhengwei
Zhang, Xiangzheng
Computation and Language
We argue that multi-document reasoning is constrained not only by how much text a model can read, but also by how limited query-time evidence budget is allocated across documents and semantic granularities. Full-context inference exposes the model to broad evidence non-selectively and at high per-query cost, while flat chunk retrieval often returns locally relevant passages that are weakly organized for cross-document synthesis. We present \textbf{BEAR}, a framework for structured evidence allocation that builds hierarchical semantic indices offline and performs coarse-to-fine evidence access at query time through complementary \emph{exploration} and \emph{recovery} paths. This coarse-to-fine design can be viewed as structured evidence allocation under a fixed evidence-context budget. Across synthetic and real-world benchmarks, BEAR performs particularly strongly on DragonBall, remains competitive with strong retrieval-based baselines on HotpotQA, and yields the best retrieval-based result on 2Wiki under our evaluated protocol, while operating under substantially smaller \emph{query-time evidence budgets} than the reported long-context references. Additional analyses suggest that the gains are associated with hierarchy as an allocation substrate together with complementary exploration and recovery, rather than semantic chunking alone.
title BEAR: Budgeted Evidence Allocation for Multi-Document Reasoning
topic Computation and Language
url https://arxiv.org/abs/2601.18116