---
title: "桥接证据：静态检索效用不能预测多步Agent搜索中的因果效用"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2607.15253v1"
permalink: "https://feed.astralor.com/items/itm_33fea838d0d5223dae30adca"
published: "2026-07-16T17:48:29.000Z"
score: 79
channel: "ai"
tags: ["Agentic搜索","HotpotQA","ReAct","因果推断","桥接证据","检索效用"]
---

# 桥接证据：静态检索效用不能预测多步Agent搜索中的因果效用

## AI 点评

> 这篇论文抓住了当下RAG和Agent结合中最关键的盲点：我们仍在用静态指标衡量动态过程。对从事搜索增强应用开发和评估的研究者来说，这篇工作直接撼动了你对“相关性”的理解，值得细读。

## 速览

论文发现当前检索系统基于静态效用（文档直接回答问题）的评估方法，在语言模型作为搜索代理时失效。通过HotpotQA上的ReAct代理实验，对23322个文档观测，静态检索效用与因果效用（CTU）几乎统计独立（Spearman ρ=-0.026）。约三分之一的文档被代理读取时因果关键，但被静态读评时无用，称为“桥接文档”。机制分析表明，桥接文档通过提供可区分实体来重定向代理后续搜索。静态相关性与因果有用性在代理检索中是不同概念，优化前者无法获得后者。

## 关键点

- 静态检索效用与因果效用（CTU）在23,322个文档观测上近乎统计独立（Spearman ρ=-0.026），说明传统评估无法反映代理场景。
- 约1/3的“桥接文档”在静态评测中无用，但对代理后续查询起因果关键作用，通过提供可区分实体重定向搜索。
- 桥接文档中的可区分实体在代理下一查询中出现频率是非相关文档实体的4.02倍（6.1% vs 1.5%）。

## 全文

Retrieval systems are trained and evaluated on a static idea of usefulness: hand a document and a question to a reader model, see whether the answer improves, and score the document accordingly. The idea holds up when a document is read on its own. It breaks when a language model works as a search agent, issuing several queries and reasoning across turns, because a document can matter for what it lets the agent do next rather than for what it says about the current question. We measure that gap rather than argue it. Using a ReAct style agent over HotpotQA, we replay 1000 development questions and, for every document the agent read, delete it and re-run the rest of the trajectory from that point. Comparing the original run against its counterfactual gives a Counterfactual Trajectory Utility (CTU) score from three deltas: final answer quality, next query retrieval quality, and turn count. Crossing CTU against Static RAG Utility (SRU) over 23,322 document observations, the two are close to statistically independent (Spearman rho = -0.026). Roughly a third of the documents the agent reads are causally load bearing while looking useless to a static reader; we call these bridge documents. The pattern survives when the reader based axis is swapped for a BM25 and cross encoder proxy, giving a bridge cell of 27.2% on an evenly spread axis. A second experiment pins down the mechanism. Using the Observable Entity Relevance (OER) measure from prior work, entities that discriminate relevant from non-relevant candidates appear in the agent's next query 4.02 times more often than entities found only in non-relevant documents (6.1% vs 1.5%, n = 227,139). A bridge document earns its keep by handing the agent a discriminative entity that redirects the search. Static relevance and causal usefulness are different quantities in agentic retrieval, and optimizing the first does not deliver the second.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_33fea838d0d5223dae30adca