Evidence Blindness in Direct Corpus Interaction: Persistent Navigation with AtlasNav

📄 arXiv: 2608.24764v1 📥 PDF

作者: Hongyu Guo, Zhiyu Zheng, Zhao Cao

分类: cs.AI

发布日期: 2026-08-25

备注: 27 pages, 7 figures. Code, data, and trajectories will be released


💡 一句话要点

提出AtlasNav以解决直接语料交互中的证据盲区问题

🎯 匹配领域: 支柱九:具身大模型 (Embodied Foundation Models)

关键词: 直接语料交互 证据盲区 AtlasNav 信息检索 智能问答 多视角导航 语料组织

📋 核心要点

  1. 现有方法在有限的交互预算下,无法有效利用可访问的证据,导致信息获取不完全。
  2. 论文提出AtlasNav框架,通过一次性组织语料库为Corpus Atlas,实现自适应导航,提升了交互效率。
  3. 实验结果显示,AtlasNav在准确率和推理成本上均优于现有动态工作空间方法,展现出更高的证据获取效率。

📝 摘要(中文)

大型语言模型代理正在超越传统的检索增强生成,向与外部语料的直接交互发展。直接语料交互(DCI)使整个语料库可访问,但在有限的交互预算下,所需证据可能无法有效呈现。我们称这种逐步的静默损失为证据盲区,并通过阶段性证据实现进行量化。我们提出AtlasNav,一个持久的多视角语料导航框架,允许每个查询自适应导航,而不是重建共享结构。在BrowseComp-Plus上,AtlasNav实现了92.05%的严格准确率,同时相较于之前的动态工作空间方法,在线推理成本降低了30.21%。

🔬 方法详解

问题定义:论文旨在解决在直接语料交互中,有限交互预算导致的证据盲区问题。现有方法在处理可访问证据时,往往无法有效呈现所需信息,造成信息获取的低效。

核心思路:AtlasNav框架的核心思路是将语料库组织为Corpus Atlas,使得每个查询能够自适应地导航,而不是依赖于动态重建共享结构,从而提高了信息检索的效率和准确性。

技术框架:AtlasNav的整体架构包括语料库的初始组织、查询的自适应导航和动态信息检索三个主要模块。首先,语料库被组织为一个结构化的Atlas;然后,针对每个查询,系统能够快速定位相关信息。

关键创新:AtlasNav的主要创新在于其持久的多视角语料导航能力,允许在有限的交互预算下实现更高效的证据获取。这与现有方法的动态重建方式形成了本质区别。

关键设计:在设计中,AtlasNav采用了特定的参数设置以优化导航效率,并通过精确的损失函数来提升模型的学习能力,确保在不同的语料组织下均能保持高效的检索性能。

🖼️ 关键图片

fig_0
img_1
img_2

📊 实验亮点

实验结果表明,AtlasNav在BrowseComp-Plus数据集上达到了92.05%的严格准确率,相较于之前的动态工作空间方法,在线推理成本降低了30.21%。在相同预算下,AtlasNav能够更早实现完整的所需证据,展现出更高的效率。

🎯 应用场景

该研究的潜在应用领域包括智能问答系统、信息检索和企业知识管理等。通过提高信息获取的效率,AtlasNav能够在实际应用中显著提升用户体验,帮助用户更快地找到所需信息,具有广泛的实际价值和未来影响。

📄 摘要(原文)

Large language model agents are moving beyond conventional retrieval-augmented generation toward direct interaction with external corpora. Direct Corpus Interaction (DCI) keeps the full corpus accessible, yet reachable evidence can remain unusable under finite interaction budgets. Required evidence may fail to surface, a surfaced supporting document may remain unopened, or an opened document may fail to expose its decisive fragment. We call this progressive silent loss Evidence Blindness and quantify it through stage-wise evidence realization. Within the DCI paradigm, raw interaction adds little reusable corpus organization, while dynamic-workspace methods reconstruct a query-conditioned interaction space from each query and trajectory. In both cases, useful structure is recovered largely online. We instead formulate large-scale agentic search as finite-budget navigation over reusable corpus structure. We introduce AtlasNav, a persistent multi-view corpus-navigation framework that retains direct corpus interaction but organizes the corpus once into a Corpus Atlas, allowing each query to navigate adaptively rather than reconstruct shared structure. On BrowseComp-Plus, AtlasNav achieves 92.05% strict accuracy while reducing recorded online inference cost by 30.21% relative to the prior dynamic-workspace state of the art. Under matched budgets, it realizes the complete required evidence earlier and approaches the same model's evidence-supplied empirical reference more rapidly. The same representation principle remains effective under PhantomWiki's distinct corpus organization and controlled 10K-1M scaling, and transfers competitively to heterogeneous enterprise knowledge. These results show that agentic search depends not only on accessible evidence, but also on how the corpus is represented so that limited interaction becomes effective navigation.