Co-adaptive Rule Routing and Constraining for Structured Knowledge Retrieval

Zhisheng Qi; Haoyu Han; Jiaxi Li; Bing He; Huiyuan Chen; Franck Dernoncourt; Ryan A. Rossi; Mahantesh M. Halappanavar

cs.IROct 10, 2026

Co-adaptive Rule Routing and Constraining for Structured Knowledge Retrieval

Authors: Zhisheng Qi, Haoyu Han, Jiaxi Li, Bing He, Huiyuan Chen, Franck Dernoncourt, Ryan A. Rossi, Mahantesh M. Halappanavar,

Organizations: University of Georgia · Michigan State University · Georgia Institute of Technology · Amazon ·

Abstract

Structured knowledge retrieval acquires task-relevant evidence from structured knowledge sources for downstream tasks. Existing methods rely on unconstrained retrieval or predefined query-rule supervision, such as author-paper rules for scientific discovery and anatomy-drug rules for precision medicine. Such supervision is rarely available in practice. We propose Co-Adaptive Rule Routing and Constraining Retrieval (CoRE), which jointly learns which reasoning rule to assign to each query and how to retrieve evidence under that rule, without predefined query-rule pairs. A routing agent selects a rule, and its corresponding retriever traverses the knowledge base and ranks evidence that satisfies the rule's constraints. Because routing and retrieval are interdependent, CoRE alternates between reinforcement learning for the routing policy and contrastive learning for the rule-constrained retriever. The selected rules provide supervision for retriever training, while retrieval performance rewards the routing policy. This closed-loop process enables both components to co-adapt without explicit query-rule supervision. Experiments on STaRK-QA datasets across scientific discovery, biomedical reasoning, and e-commerce knowledge management demonstrate CoRE's effectiveness. Ablations show the benefit of co-adapting the router and retriever. Generation-based routing transfers across datasets and benefits from multi-domain pretraining followed by target-specific fine-tuning. Analysis of rule usage across queries also informs automatic rule refinement and discovery.

Figures & tables

Figure 1: (a) Given a patient’s biomedical query, structured knowledge retrieval is guided by the reasoning rule $\mathtt{Anatomy\rightarrow Disease}$ , which a doctor gradually acquires through years of clinical experience. (b) Co-adapting the rule router and rule-constraining retriever further improves retrieval performance.

Figure 2: Overview of the CoRE framework. CoRE alternates between optimizing rule-constraining retrievers and the rule-routing agent. Given the current query–rule assignments, each retriever is first trained on queries assigned to its corresponding rule (①). The resulting retrieval performance is converted into rule-level reward signals (②), which are used to update the routing policy (③). The updated router then reassigns queries to rules (④), providing new supervision for the next round of retriever optimization. This alternating process enables the routing policy and rule-constraining retrievers to co-adapt through a closed feedback loop.

Category	Retrieval Baseline	AMAZON				MAG				PRIME				AVERAGE
Category	Retrieval Baseline	H@1	H@5	R@20	MRR	H@1	H@5	R@20	MRR	H@1	H@5	R@20	MRR	H@1	H@5	R@20	MRR
Textual	BM25	40.62	62.18	50.89	50.59	23.11	44.47	44.02	33.23	11.96	25.67	28.31	18.35	25.23	44.11	41.07	34.06
	Ada	39.16	62.73	53.29	50.35	29.08	49.61	48.36	38.62	12.63	31.49	36.00	21.41	26.96	47.94	45.88	36.79
	Multi-Ada	40.07	64.98	55.12	51.55	25.92	50.43	50.80	36.94	15.10	33.56	38.05	23.49	27.03	49.66	47.99	37.33
	DPR	15.29	47.93	44.49	30.20	10.51	35.23	42.11	21.34	4.46	21.85	30.13	12.38	10.09	35.00	38.91	21.31
Graph	QAGNN	26.56	50.01	52.05	37.75	12.88	39.01	46.97	29.12	8.85	21.35	29.63	14.73	16.10	36.79	42.88	27.20

Table 1: Retrieval performance on AMAZON, MAG, and PRIME, comparing baselines with CoRE-BERT-C , CoRE-LLM-C , and CoRE-LLM-G . Best and second-best results are bold and underlined , respectively.

Method	AMAZON		MAG		PRIME		Average
Method	H@1	MRR	H@1	MRR	H@1	MRR	H@1	MRR
CoRE-BERT-C	50.71	62.46	63.37	74.17	49.34	61.27	54.47	65.97
w/o Agent-Adapt (unadapted BERT)	14.63	19.83	28.23	35.05	5.15	8.15	16.00	21.01
w/o Agent-Adapt (GPT-5-mini)	26.70	32.72	28.23	31.87	4.14	6.66	19.69	23.75
w/o Retriever-Adapt (BM25-traversal)	49.87	61.45	55.51	65.45	38.86	49.70	48.08	58.87

Table 2: Ablation of routing agent and retriever adaptation. w/o Agent-Adapt fixes the router to a pre-trained or zero-shot model while continuing retriever training. w/o Retriever-Adapt replaces learned rule-constraining retrievers with fixed BM25-traversal scoring while continuing routing agent training. Best results are in bold .

Figure 3: Cross-dataset H@1 of CoRE-LLM-G. Parentheses show gains over zero-shot Qwen with trained retrievers.

Figure 4: H@1 of CoRE-LLM-G under different training using Qwen2.5-32B-Instruct as the base router. Multi-domain pretraining followed by target-specific fine-tuning performs best.

Figure 5: Optimization dynamics of CoRE-BERT-C on MAG. Each round alternates between retriever and router adaptation; the insets show the routing distributions before training and after Rounds 1 and 3.

Appendix figures & tables10 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 6: Rule-wise fractional oracle frequency across MAG, Amazon, and PRIME. Each stacked bar decomposes a rule’s oracle credit into uniquely optimal ( Unique ) and tied-optimal ( Tied ) cases. For a $k$ -way tie, each optimal rule receives $1/k$ credit, so the frequencies sum to 100% within each dataset.

Figure 7: Case study of GPT-5 retrieval-rule routing failures on MAG. (a) Confusion matrix between GPT-5-selected rules and highest-reward oracle rules. (b) Semantic–retriever misalignment : GPT-5 selects the semantically plausible iapf rule, but the added field constraint is misgrounded by the retriever and prevents access to the ground-truth paper, while the broader iap path succeeds. (c) Semantic mis-grounding : GPT-5 treats the explicit field “Seiberg–Witten theory” as free-text content and selects p , whereas fp correctly grounds the field and narrows retrieval to structurally relevant papers.

Figure 8: Comparison of training strategies for the BERT-C router, averaged over AMAZON, MAG, and PRIME. RLFT consistently outperforms hard-label and soft-label SFT across all retrieval metrics.

Figure 9: Performance with a fixed trained retriever and different routers, averaged over AMAZON, MAG, and PRIME. Stronger zero-shot LLM routers improve retrieval, but CoRE-LLM-G remains best across all metrics, showing the benefit of retrieval-feedback-based router adaptation.

Figure 10: Average Hit@1 across datasets under different router–retriever combinations. Rule-constraining retrieval consistently improves over unconstrained BM25, while stronger retrievers provide additional gains within the selected rule space. CoRE-LLM-G remains best overall.

Figure 11: Efficiency comparison on MAG in terms of mean inference time per question. We decompose the cost into routing/LLM generation and retrieval/tool execution. CoRE substantially reduces inference cost compared with step-wise agentic graph exploration, particularly when using classifier-based routing. The y-axis is shown on a logarithmic scale.

	STaRK-Amazon	STaRK-MAG	STaRK-Prime
#entity types	4	4	10
#relation types	5	4	18
#entities	1,035,542	1,872,968	129,375
#relations	9,443,802	39,802,116	8,100,498
#tokens	592,067,882	212,602,571	31,844,769
#queries	9,181	13,407	11,302

Appendix

Table 3: Statistics of the STaRK datasets, and generation-based routing configuration.

Key	Traversal Path
A	Product
B	Brand $\mathrel{\;\rightarrow\;}$ Product
C	Product $\mathrel{\;\rightarrow\;}$ Product
D	Category $\mathrel{\;\rightarrow\;}$ Product
E	Product $\mathrel{\;\rightarrow\;}$ Category $\mathrel{\;\rightarrow\;}$ Product
F	Product $\mathrel{\;\rightarrow\;}$ Brand $\mathrel{\;\rightarrow\;}$ Product

Appendix

Table 4: Candidate retrieval rules for Amazon.

Key	Traversal Path
A	Paper
B	Author $\mathrel{\;\rightarrow\;}$ Paper
C	Paper $\mathrel{\;\rightarrow\;}$ Paper
D	Field of Study $\mathrel{\;\rightarrow\;}$ Paper
E	Institution $\mathrel{\;\rightarrow\;}$ Author $\mathrel{\;\rightarrow\;}$ Paper
F	Paper $\mathrel{\;\rightarrow\;}$ Author $\mathrel{\;\rightarrow\;}$ Paper

Appendix

Table 5: Candidate retrieval rules for MAG.

Key	Traversal Path
R00	Anatomy $\mathrel{\;\rightarrow\;}$ Expression Present $\mathrel{\;\rightarrow\;}$ Gene/Protein
R01	Anatomy $\mathrel{\;\rightarrow\;}$ Expression Absent $\mathrel{\;\rightarrow\;}$ Gene/Protein
R02	Pathway $\mathrel{\;\rightarrow\;}$ Interacts With $\mathrel{\;\rightarrow\;}$ Gene/Protein $\mathrel{\;\rightarrow\;}$ Expression Absent $\mathrel{\;\rightarrow\;}$ Anatomy
R03	Gene/Protein $\mathrel{\;\rightarrow\;}$ Target $\mathrel{\;\rightarrow\;}$ Drug
R04	Disease $\mathrel{\;\rightarrow\;}$ Associated With $\mathrel{\;\rightarrow\;}$ Gene/Protein
R05	Gene/Protein $\mathrel{\;\rightarrow\;}$ PPI $\mathrel{\;\rightarrow\;}$ Gene/Protein

Appendix

Table 6: Candidate retrieval rules for PRIME.

Information Retrieval

Information Retrieval studies how to find, rank, and organize relevant information in large collections of data.

See all 474 papers

Beyond the Query: Do Retrieval Signals Improve Adaptive Multimodal RAG Routing?
Sep 11, 2026Qiaomu Li, Qiuyuan Zhang, Nong MingMultimodal RAG Adaptive Model Routing
College of Computing and Software Engineering Kennesaw State University Marietta, GA 30060, USA
ORDER: Task-Conditioned Routing for Retrieval-Augmented Generation
Sep 15, 2026Aurélien Pellet, Julien Perez, Marie PurenRetrieval-Augmented Generation Adaptive RAG
LRE · LRE, EPITA · EPITECH +2
It Takes Two to Match: Co-Evolving Generative Retriever with Reinforcement Learning
Sep 1, 2026Runpeng Dai, Kaili Huang, Changsung Kang +1Reinforcement Learning Generative Retrieval
University of North Carolina at Chapel Hill · Apple

Sep 11, 2026J/K move · Enter open · S save

Beyond the Query: Do Retrieval Signals Improve Adaptive Multimodal RAG Routing?

Qiaomu Li, Qiuyuan Zhang, Nong Ming

College of Computing and Software Engineering Kennesaw State University Marietta, GA 30060, USA

Multimodal RAG Adaptive Model Routing

Open paper AskPDF arXiv

Adaptive RAG often uses retrieval-time signals to decide whether another retrieval, reranking, or multimodal step should run. We ask whether these signals add routing value once the query itself is already known. Across document, audio, and video RAG, we compare matched query-only and query+retrieval routers while holding the optional actions, router family, training procedure, and evaluation fixed. On the held-out final evaluation, adding the tested retrieval signals does not produce a reliable routing improvement over the query-only baseline. Some retrieval signals are associated with whether a later step will help, but that predictability does not consistently lead to bet- ter RUN/SKIP decisions. The main lesson is therefore methodological: retrieval-state features should not be credited with routing value unless they improve over a matched query-only control. Our results do not show that routing or retrieval state is generally useless; they show that the incremental value of retrieval signals must be demonstrated rather than assumed.

Abstract

Information Retrieval

Explore similar work

Beyond the Query: Do Retrieval Signals Improve Adaptive Multimodal RAG Routing?

ORDER: Task-Conditioned Routing for Retrieval-Augmented Generation

It Takes Two to Match: Co-Evolving Generative Retriever with Reinforcement Learning