--- title: "AgentRouter — paper by Zhang et al. (2026)" description: "With so many LLMs and agent strategies available, it is hard to know which configuration will… 2% F1. 2 references, cited by 3 papers here." authors: ["Zheyuan Zhang", "Kaiwen Shi", "Zhengqing Yuan", "Zehong Wang", "Tianyi Ma", "Keerthiram Murugesan", "Vincent Galassi", "Chuxu Zhang", "Yanfang Ye"] year: 2026 venue: "Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)" source: "https://aclanthology.org/2026.acl-long.33/" doi: "10.18653/v1/2026.acl-long.33" licence: "CC BY 4.0 (ACL Anthology)" licence_url: "https://creativecommons.org/licenses/by/4.0/" summary_licence: "Elamur Cosmic License" canonical: "https://research.elamur.ai/free/en/papers/zhang-2026-agentrouter-knowledge-graph-guided-llm/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/zhang-2026-agentrouter-knowledge-graph-guided-llm/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/4/) / AgentRouter: A Knowledge-Graph-Guided LLM Router for Collaborative Multi-Agent Question Answering # AgentRouter: A Knowledge-Graph-Guided LLM Router for Collaborative Multi-Agent Question Answering **Authors:** [Zheyuan Zhang](https://md.research.elamur.ai/free/en/authors/zheyuan-zhang/), [Kaiwen Shi](https://md.research.elamur.ai/free/en/authors/kaiwen-shi/), [Zhengqing Yuan](https://md.research.elamur.ai/free/en/authors/zhengqing-yuan/), [Zehong Wang](https://md.research.elamur.ai/free/en/authors/zehong-wang/), [Tianyi Ma](https://md.research.elamur.ai/free/en/authors/tianyi-ma/), [Keerthiram Murugesan](https://md.research.elamur.ai/free/en/authors/keerthiram-murugesan/), [Vincent Galassi](https://md.research.elamur.ai/free/en/authors/vincent-galassi/), [Chuxu Zhang](https://md.research.elamur.ai/free/en/authors/chuxu-zhang/) and [Yanfang Ye](https://md.research.elamur.ai/free/en/authors/yanfang-ye/) · **Year:** 2026 · **Venue:** [Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)](https://md.research.elamur.ai/free/en/venues/acl/) ([ACL 2026](https://md.research.elamur.ai/free/en/venues/acl/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/) **Links:** [Publisher's page](https://aclanthology.org/2026.acl-long.33/) · [PDF](https://aclanthology.org/2026.acl-long.33.pdf) · [DOI 10.18653/v1/2026.acl-long.33](https://doi.org/10.18653/v1/2026.acl-long.33) · [OpenAlex](https://openalex.org/W4414977016) **Citations:** 1 citations (OpenAlex, 2026-10-14) · [3](https://md.research.elamur.ai/free/en/papers/zhang-2026-agentrouter-knowledge-graph-guided-llm/cited-by/) in our library, of the 149,439 papers whose bibliographies we read ## Summary With so many LLMs and agent strategies available, it is hard to know which configuration will work best for a given QA task. Agents on the same backbone can differ widely in quality, and their rankings change across datasets. AgentRouter addresses this by treating multi-agent QA as a routing problem over a knowledge graph. Each QA instance becomes a heterogeneous graph that links the query, entities extracted from the context with spaCy, and a pool of 24 agents. A type-aware heterogeneous GNN, called RouterGNN, passes messages across these node types and outputs a weight distribution over agents. It is trained with KL divergence against each agent's measured performance rather than against a single "best agent" label. The final answer combines agent outputs by weighted voting, optionally limited to the top-K agents. Tested on 2Wiki, HotpotQA, NewsQA and TriviaQA, the router beats the best single agent and ensemble baselines, with the largest gains on multi-hop data. This matters because it shows that learned, context-aware routing can outperform heuristic voting or LLM-based judges. ## Key points - **Motivating evidence:** Six classic agent designs (raw, CoT, self-consistency, MAD, ReAct-Reflect, summary) were run on Llama-3-8B-Instruct and Mixtral-8×7B-Instruct over HotpotQA and NewsQA. Their F1 distributions were wide and their rankings unstable across tasks and backbones. - **Graph size:** Each instance graph contains 1 query node and 24 agent nodes. On average it has about 111.6 entity nodes for HotpotQA and 46.0 for NewsQA, and HotpotQA graphs are much denser in entity–entity edges. - **Top-K ablation:** On multi-hop datasets, keeping a moderate number of agents (roughly 5–10) works best. On single-hop QA, including more agents tends to give better and steadier results. - **Hyperparameters:** Making the GNN deeper does not reliably raise F1. Larger hidden dimensions give more consistent improvements. - **Cross-dataset transfer is asymmetric:** - Training on 2Wiki and testing on HotpotQA costs at most about 2% F1. - Training on TriviaQA and testing on 2Wiki loses roughly 29–38% F1. - This suggests the router needs training data from the target task. *Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.* ## Institutions - [IBM](https://md.research.elamur.ai/free/en/companies/ibm/) — Company, Armonk, United States · 154 works · 72,606 citations - [University of Notre Dame](https://md.research.elamur.ai/free/en/universities/university-of-notre-dame/) — University, Notre Dame, United States · 59 works · 37,127 citations - [University of Connecticut](https://md.research.elamur.ai/free/en/universities/university-of-connecticut/) — University, Storrs, United States · 13 works · 357 citations 📚 Shelves (5): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works · [👁️ Computer Vision and Pattern Recognition](https://md.research.elamur.ai/free/en/shelves/computer-science/computer-vision-and-pattern-recognition/) · 3,486 works 🧭 Topics (2): [❓ Question answering](https://md.research.elamur.ai/free/en/topics/question-answering/) · 487 works · [🕸️ Knowledge graphs & ontologies](https://md.research.elamur.ai/free/en/topics/knowledge-graphs/) · 211 works ## Cite Cite as: Zheyuan Zhang, Kaiwen Shi, Zhengqing Yuan, Zehong Wang, Tianyi Ma, Keerthiram Murugesan, Vincent Galassi, Chuxu Zhang and Yanfang Ye. 2026. AgentRouter: A Knowledge-Graph-Guided LLM Router for Collaborative Multi-Agent Question Answering. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @inproceedings{zhang2026agentrouter, title = {{AgentRouter: A Knowledge-Graph-Guided LLM Router for Collaborative Multi-Agent Question Answering}}, author = {Zheyuan Zhang, Kaiwen Shi, Zhengqing Yuan, Zehong Wang, Tianyi Ma, Keerthiram Murugesan, Vincent Galassi, Chuxu Zhang and Yanfang Ye}, booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)}, year = {2026}, doi = {10.18653/v1/2026.acl-long.33}, url = {https://doi.org/10.18653/v1/2026.acl-long.33} } ``` - [← previous AgencyBench: Benchmarking the Frontiers of Autonomous Agents in 1M-Token Real-World Contexts](https://md.research.elamur.ai/free/en/papers/li-2026-agencybench-benchmarking-frontiers-autonomous-agents/) - [next → Aleph-Alpha-GermanWeb: Improving German-language LLM pre-training with model-based data curation and synthetic data generation](https://md.research.elamur.ai/free/en/papers/burns-2026-aleph-alpha-germanweb-improving-german/) - [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/4/) ## References 2 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. This work was partially supported by the NSF under grants IIS-2533550, IIS-2321504, IIS-2217239, CNS-2426514, and CMMI-2146076, Notre Dame Strategic Framework Research Grant (2025), and Notre Dame Poverty Research Package (2025). Any expressed opinions, findings, and conclusions or recommendations a… [find](https://scholar.google.com/scholar?q=This+work+was+partially+supported+by+the+NSF+under+grants+IIS%2D2533550%2C+IIS%2D2321504%2C+IIS%2D2217239%2C+CNS%2D2426514%2C+and+CMMI%2D2146076%2C+Notre+Dame+Strategic+Framework+Research+Grant+%282025%29%2C+and+Notre+Dame+Pov) 2. Qwen Alibaba. 2025. Qwen2.5-7b-instruct. [https://huggingface.co/Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct). \#\# Limitations [↗](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) *Updated 1 November 2026*