A New Role for Relevance: Guiding Corpus Interaction in Agentic Search
2026-07-27 • Computation and Language
Computation and Language
AI summaryⓘ
The authors explain that finding exactly useful information in a large set of documents is tricky because just knowing if a document is relevant isn't enough to locate precise answers. They propose a new method called RARG that uses relevance information to guide a detailed search, helping the system focus on the most promising parts of the documents early on. This approach lets the search be more efficient and accurate when answering complex questions. Their tests show that RARG works better than previous methods that either only rank documents or search blindly.
relevancedocument retrievaldirect corpus interactionripgrepsearch agentquery relevanceinformation retrievallarge language modelsbrowse question answeringreranking
Authors
Jiangnan Li, Yuqing Li, Mo Yu, Jinchao Zhang, Jie Zhou
Abstract
Relevance is a query-dependent estimate of whether a document or excerpt contains useful evidence. Existing retrieval agents use relevance to select top-$k$ content, but document relevance alone cannot localize, compose, or verify the evidence required by complex questions. Direct Corpus Interaction (DCI) enables such fine-grained operations through grep-style exploration, but its relevance-agnostic search can expose useful clues late and delay convergence. Recent advances use relevance to narrow the corpus into a working space for interaction. Once interaction begins, however, relevance still does not directly guide which documents grep searches first or distinguish informative excerpts from a broad set of matches to let LLMs see them first. We introduce the Relevance-Aware RipGrep Search Agent (RARG), which turns relevance into an execution prior for corpus interaction. RARG provides coarse-to-fine relevance guidance: it orders documents for sequential 'ripgrep' traversal to expose globally relevant clues earlier, initializes promising entry points with query-relevant paragraphs, and reranks grep matches to surface informative excerpts that document-level ranking may otherwise obscure. Across challenging browse question answering and reasoning-intensive retrieval, RARG improves the accuracy--efficiency frontier over retrieval-based and direct-interaction agents. These results demonstrate that relevance-aware interaction enables faster and more reliable search convergence.