StateRAG：为复杂检索增强生成引入类型化状态契约

View PDF HTML (experimental)

Abstract:Multi-step retrieval-augmented generation requires tracking evolving information needs, selected evidence, and unresolved gaps. However, free-form trajectories conflate this history, forcing each step to implicitly reconstruct the current state, while existing explicit-state methods remain vulnerable to inconsistent state overwrites. To address these limitations, we introduce StateRAG, a framework that coordinates a Planner, Navigator, Retriever, and Verifier through an explicit retrieval state with checked updates. These roles propose updates across four state fields: the plan, retrieval trace, evidence, and sufficiency assessment. A dedicated controller validates each proposal against the accepted state, rejecting inconsistent updates and guiding revisions within the query budget. A final Reader then answers strictly from the validated evidence. Ablation experiments on a shared graph backend show that StateRAG outperforms free-form trajectory and unverified explicit-state baselines on LongBench by 7.7 and 4.1 Macro F1 points, respectively. These gains consistently extend to document QA and flat passage retrieval. Furthermore, StateRAG surpasses the strongest external baselines by 5.3, 5.0, and 4.3 points on LongBench, QASPER, and DocVQA.

Subjects:	Computation and Language (cs.CL)
Cite as:	arXiv:2605.25379 [cs.CL]
	(or arXiv:2605.25379v4 [cs.CL] for this version)
	https://doi.org/10.48550/arXiv.2605.25379 arXiv-issued DOI via DataCite

Submission history

From: Miaohe Niu [view email]
[v1] Mon, 25 May 2026 03:08:53 UTC (12,692 KB)
[v2] Wed, 29 Jul 2026 12:34:25 UTC (647 KB)
[v3] Mon, 9 Nov 2026 02:31:19 UTC (1,497 KB)
[v4] Tue, 10 Nov 2026 13:00:37 UTC (1,497 KB)