--- title: "DFAMS — paper by Yang et al. (2026) — summary and ideas" description: "Federated Retrieval (FR) sends a query to several separate knowledge bases instead of one central index. 85.03 accuracy. 2 references." authors: ["Zhibang Yang", "Xinke Jiang", "Rihong Qiu", "Ruiqing Li", "Yihang Zhang", "Yue Fang", "Yongxin Xu", "Hongxin Ding", "Xu Chu", "Junfeng Zhao", "Yasha Wang"] 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.1135/" doi: "10.18653/v1/2026.acl-long.1135" 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/yang-2026-dfams-dynamic-flow-guided-federated/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/yang-2026-dfams-dynamic-flow-guided-federated/) [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/29/) / DFAMS: Dynamic-flow guided Federated Alignment based Multi-prototype Search # DFAMS: Dynamic-flow guided Federated Alignment based Multi-prototype Search **Authors:** Zhibang Yang, [Xinke Jiang](https://md.research.elamur.ai/free/en/authors/xinke-jiang/), [Rihong Qiu](https://md.research.elamur.ai/free/en/authors/rihong-qiu/), Ruiqing Li, Yihang Zhang, [Yue Fang](https://md.research.elamur.ai/free/en/authors/yue-fang/), [Yongxin Xu](https://md.research.elamur.ai/free/en/authors/yongxin-xu/), [Hongxin Ding](https://md.research.elamur.ai/free/en/authors/hongxin-ding/), [Xu Chu](https://md.research.elamur.ai/free/en/authors/xu-chu/), [Junfeng Zhao](https://md.research.elamur.ai/free/en/authors/junfeng-zhao/) and [Yasha Wang](https://md.research.elamur.ai/free/en/authors/yasha-wang/) · **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.1135/) · [PDF](https://aclanthology.org/2026.acl-long.1135.pdf) · [DOI 10.18653/v1/2026.acl-long.1135](https://doi.org/10.18653/v1/2026.acl-long.1135) · [OpenAlex](https://openalex.org/W4414447943) **Citations:** 0 citations (OpenAlex, 2026-10-14) ## Summary Federated Retrieval (FR) sends a query to several separate knowledge bases instead of one central index. Current FR systems often pick the wrong source or return weak documents when a query is ambiguous or crosses domains. DFAMS addresses this using the Dynamic Information Flow (DIF), meaning the activation paths that form inside an LLM during inference. The method works in two stages: - **Probing:** gradients from a small set of annotated queries feed a Shapley value-based attribution step, which finds the neurons tied to user intent and subdomain boundaries. - **Alignment and routing:** the extracted DIF features train an Align-MLP with multi-prototype contrastive learning. This models subdomains within each source and aligns related content across sources. At inference, Adaptive Prototype-Guided Routing decides whether retrieval is needed and which sources to query. The authors also built a harder FR benchmark. On five benchmarks, DFAMS beats prior FR methods in source classification, recall and QA accuracy. This matters wherever knowledge is split across silos that cannot simply be merged, such as medical records, textbooks and papers. ## Key points - Reported gains over strong FR baselines reach up to +14.37% knowledge classification accuracy, +5.38% retrieval recall and +6.45% downstream QA accuracy. - Shapley heatmaps were computed for Qwen2.5-7B, LLaMA3.1-8B, Qwen2.5-3B and Qwen2.5-0.5B. They show two peaks: shallow layers carry intent signals and deep layers carry domain signals, so DIF is extracted from the deep layers. - A 2000-dim DIF representation outperforms both random features of the same size and full-layer 3584-dim features, even without training. On Wiki the gain is +14.6% accuracy. - In the Wiki ablation, the full model reaches 85.03 accuracy and 53.83 recall. Removing inter-KB alignment lowers accuracy by 9.14 points, and removing Semantic Routing lowers recall by 5.16 points. Three prototypes per class and top-T = 3 work best. - The new benchmark covers knowledge-free queries, multi-fragment queries within one source, and cross-source queries. Code is available at github.com/Artessay/DFAMS. *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 - [Peking University](https://md.research.elamur.ai/free/en/universities/peking-university/) — University, Beijing, China · 315 works · 43,267 citations 📚 Shelves (4): [🖥️ 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 ## Cite Cite as: Zhibang Yang, Xinke Jiang, Rihong Qiu, Ruiqing Li, Yihang Zhang, Yue Fang, Yongxin Xu, Hongxin Ding, Xu Chu, Junfeng Zhao and Yasha Wang. 2026. DFAMS: Dynamic-flow guided Federated Alignment based Multi-prototype Search. 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{yang2026dfams, title = {{DFAMS: Dynamic-flow guided Federated Alignment based Multi-prototype Search}}, author = {Zhibang Yang, Xinke Jiang, Rihong Qiu, Ruiqing Li, Yihang Zhang, Yue Fang, Yongxin Xu, Hongxin Ding, Xu Chu, Junfeng Zhao and Yasha Wang}, 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.1135}, url = {https://doi.org/10.18653/v1/2026.acl-long.1135} } ``` - [← previous DevLake at LoResMT 2026: The Impact of Pre-training and Model Scale on Russian-Bashkir Low-Resource Translation](https://md.research.elamur.ai/free/en/papers/tyurin-2026-devlake-loresmt-2026-impact-pre/) - [next → DIA-HARM: Dialectal Disparities in Harmful Content Detection Across 50 English Dialects](https://md.research.elamur.ai/free/en/papers/lucas-2026-dia-harm-dialectal-disparities-harmful/) - [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/29/) ## 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 research was supported by the Independent Research Project of the National Key Laboratory of Big Data and Decision, the New Generation Artificial Intelligence-National Science and Technology Major Project (Grant No. 2025ZD0122904), and the National Natural Science Foundation of China (Grant No.… [find](https://scholar.google.com/scholar?q=This+research+was+supported+by+the+Independent+Research+Project+of+the+National+Key+Laboratory+of+Big+Data+and+Decision%2C+the+New+Generation+Artificial+Intelligence%2DNational+Science+and+Technology+Majo) 2. Kamil Adamczewski, Yawei Li, and Luc van Gool. 2024. Shapley pruning for neural network compression. Preprint, arXiv:2407.15875. \#\# Limitations — [arXiv:2407.15875](https://arxiv.org/abs/2407.15875) [find](https://scholar.google.com/scholar?q=Kamil+Adamczewski%2C+Yawei+Li%2C+and+Luc+van+Gool%2E+2024%2E+Shapley+pruning+for+neural+network+compression%2E+Preprint%2C+arXiv%3A2407%2E15875%2E+%5C%23%5C%23+Limitations+%E2%80%94+%5BarXiv%3A2407%2E15875%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2407%2E15875%29) *Updated 2 November 2026*