Elamur Research⌕
☰
home / papers / 2026 / ARK: Answer-Centric Retriever Tuning via KG-augmented Curriculum Learning

ARK: Answer-Centric Retriever Tuning via KG-augmented Curriculum Learning

Paper · 2026 · Jiawei Zhou, Hang Ding and Haiyun Jiang

Summary

ARK addresses a weakness of retrievers in long-context RAG. They are trained to match queries to similar text, so they often miss sparse passages that actually let a generator produce the right answer. The authors propose a fine-tuning recipe that targets answer usefulness instead of surface similarity.

Positive chunks are chosen with an in-context sufficiency metric that combines three alignment signals: Forward, Backward, and Retriever/Parameter. Hard negatives come from an LLM-built knowledge graph. Personalized PageRank (PPR) is seeded on answer entities to extract query-specific subgraphs, which then guide an LLM to write augmented queries. Chunks those queries retrieve, but which are not answer-sufficient, become distractors. A three-stage contrastive curriculum moves from in-batch negatives to coarse and then fine-grained hard negatives.

On 10 datasets from UltraDomain and LongBench, the tuned retriever leads on 8. The KG is only needed during training, and the architecture is unchanged. This makes ARK a cheap drop-in upgrade compared with graph-heavy systems like GraphRAG or HippoRAG.

Key points

  • The base encoder is Qwen3-Embedding. Average F1 rises by 14.5% over the base model, with top results on 8 of 10 UltraDomain/LongBench datasets.
  • ARK was trained on Finance and Legal data and generalizes to unseen domains. Gains are largest on multi-hop sets, for example MuSiQue F1 rising from 14.19 to 20.60 and HotpotQA from 35.27 to 42.35. HippoRAG stays competitive on 2Wiki.
  • Comparisons include GraphRAG, LightRAG, HippoRAG, MemoRAG, BGE-M3, Stella, Jina and Qwen3 embeddings, using F1 and pairwise win rates.
  • In the ablation, removing Forward alignment hurts most. Each curriculum stage adds further improvement.
  • The retriever transfers to Llama-3.1-8B-Instruct and Qwen2.5-7B-Instruct, with one small drop on NarrativeQA with Llama. Code is available at github.com/valleysprings/ARK.

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

🎓Shanghai Jiao Tong UniversityShanghai · China212 works · 19,544 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksComputer Vision and Pattern Recognition · 3,486 worksInformation Systems · 1,730 works

🧭 Topics 1Research subjects and languages the work is about.Information retrieval & search · 552 works

Cite

Cite as Jiawei Zhou, Hang Ding and Haiyun Jiang. 2026. ARK: Answer-Centric Retriever Tuning via KG-augmented Curriculum Learning. 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. DOI 10.18653/v1/2026.acl-long.371

@inproceedings{zhou2026ark,
  title = {{ARK: Answer-Centric Retriever Tuning via KG-augmented Curriculum Learning}},
  author = {Jiawei Zhou, Hang Ding and Haiyun Jiang},
  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.371},
  url = {https://doi.org/10.18653/v1/2026.acl-long.371}
}

References

42 references, 23 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, and 1 others. 2023. Gpt-4 technical report. arXiv preprint arXiv:2303.08774.
  2. Reid Andersen and Fan Chung. 2007. Detecting sharp drops in pagerank and a simplified local partitioning algorithm. In International Conference on Theory and Applications of Models of Computation, pages 1–12. Springer. find
  3. Yushi Bai, Xin Lv, Jiajie Zhang, Hongchang Lyu, Jiankai Tang, Zhidian Huang, Zhengxiao Du, Xiao Liu, Aohan Zeng, Lei Hou, and 1 others. 2023. Longbench: A bilingual, multitask benchmark for long context understanding. arXiv preprint arXiv:2308.14508.
  4. Sebastian Borgeaud, Arthur Mensch, Jordan Hoffmann, Trevor Cai, Eliza Rutherford, Katie Millican, George Bm Van Den Driessche, Jean-Baptiste Lespiau, Bogdan Damoc, Aidan Clark, and 1 others. 2022. Improving language models by retrieving from trillions of tokens. In International conference on machin… find
  5. Chi-Min Chan, Chunpu Xu, Ruibin Yuan, Hongyin Luo, Wei Xue, Yike Guo, and Jie Fu. 2024. Rq-rag: Learning to refine queries for retrieval augmented generation. arXiv preprint arXiv:2404.00610.
  6. Ge Chang, Jinbo Su, Jiacheng Liu, Pengfei Yang, Yuhao Shang, Huiwen Zheng, Hongli Ma, Yan Liang, Yuanchun Li, and Yunxin Liu. 2025. Grail: Learning to interact with large knowledge graphs for retrieval augmented reasoning. arXiv preprint arXiv:2508.05498. — [arXiv:2508.05498](https://arxiv.org/abs/2508.05498) find
  7. Jianlv Chen, Shitao Xiao, Peitian Zhang, Kun Luo, Defu Lian, and Zheng Liu. 2023. Bge m3-embedding: Multi-lingual, multi-functionality, multi-granularity text embeddings through self-knowledge distillation. Preprint, arXiv:2309.07597.
  8. Albert Q Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, and 1 others. 2023. Mistral 7b. arXiv preprint arXiv:2310.06825. find
  9. Siyuan Chen, Hang Ding, Kangxiaoyu, and Jiechao Gao. 2026. GRO-RAG: Gradient-aware re-rank optimization for multi-source retrieval-augmented generation. In The Fourteenth International Conference on Learning Representations. find
  10. Vladimir Karpukhin, Barlas Oguz, Sewon Min, Patrick SH Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih. 2020. Dense passage retrieval for open-domain question answering. In EMNLP, pages 6769–6781.
  11. Gheorghe Comanici, Eric Bieber, Mike Schaekermann, Ice Pasupat, Noveen Sachdeva, Inderjit Dhillon, Marcel Blistein, Ori Ram, Dan Zhang, Evan Rosen, and 1 others. 2025. Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities. arX…
  12. Tomáš Kočiskỳ, Jonathan Schwarz, Phil Blunsom, Chris Dyer, Karl Moritz Hermann, Gábor Melis, and Edward Grefenstette. 2018. The narrativeqa reading comprehension challenge. Transactions of the Association for Computational Linguistics, 6:317–328.
  13. Pradeep Dasigi, Kyle Lo, Iz Beltagy, Arman Cohan, Noah A Smith, and Matt Gardner. 2021. A dataset of information-seeking questions and answers anchored in research papers. arXiv preprint arXiv:2105.03011.
  14. Hang Ding, Yilun Zhao, Tiansheng Hu, Manasi Patwardhan, and Arman Cohan. 2026. SciRAG: Adaptive, citation-aware, and outline-guided retrieval and synthesis for scientific literature. In Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volum…
  15. Darren Edge, Ha Trinh, Newman Cheng, Joshua Bradley, Alex Chao, Apurva Mody, Steven Truitt, and Jonathan Larson. 2024. From local to global: A graph rag approach to query-focused summarization. arXiv preprint arXiv:2404.16130. find
  16. Luyu Gao, Xueguang Ma, Jimmy Lin, and Jamie Callan. 2022. Precise zero-shot dense retrieval without relevance labels. Preprint, arXiv:2212.10496.
  17. Zirui Guo, Lianghao Xia, Yanhua Yu, Tu Ao, and Chao Huang. 2025. Lightrag: Simple and fast retrieval-augmented generation. Preprint, arXiv:2410.05779.
  18. Bernal Jiménez Gutiérrez, Yiheng Shu, Yu Gu, Michihiro Yasunaga, and Yu Su. 2024. Hipporag: Neurobiologically inspired long-term memory for large language models. In The Thirty-eighth Annual Conference on Neural Information Processing Systems.
  19. Angeliki Lazaridou, Adhi Kuncoro, Elena Gribovskaya, Devang Agrawal, Adam Liska, Tayfun Terzi, Mai Gimenez, Cyprien de Masson d’Autume, Tomas Kocisky, Sebastian Ruder, and 1 others. 2021. Mind the gap: Assessing temporal generalization in neural language models. Advances in Neural Information Proces… find
  20. Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, and 1 others. 2020. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in Neural Information Processing Systems, 33:9459–…
  21. Jian Li, Yizhang Jin, Dongqi Liu, Hang Ding, Jiafu Wu, Dongsheng Chen, Yunhang Shen, Yulei Qin, Ying Tai, Chengjie Wang, Xiaotong Yuan, and Yabiao Wang. 2026. Se-search: Self-evolving search agent via memory and dense reward. Preprint, arXiv:2603.03293. — [arXiv:2603.03293](https://arxiv.org/abs/2603.03293) find
  22. Tianle Li, Ge Zhang, Quy Duc Do, Xiang Yue, and Wenhu Chen. 2024. Long-context llms struggle with long in-context learning. arXiv preprint arXiv:2404.02060.
  23. Dongqi Liu, Hang Ding, Qiming Feng, Jian Li, Xurong Xie, Zhucun Xue, Chengjie Wang, Jiangning Zhang, and Yabiao Wang. 2026. Disco-rag: Discourse-aware retrieval-augmented generation. Preprint, arXiv:2601.04377. find
  24. Taher H Haveliwala and Sepandar D Kamvar. 2003. The second eigenvalue of the google matrix. Technical report, Citeseer. find
  25. Niklas Muennighoff, Nouamane Tazi, Loïc Magne, and Nils Reimers. 2022. Mteb: Massive text embedding benchmark. arXiv preprint arXiv:2210.07316.
  26. Xanh Ho, Anh-Khoa Duong Nguyen, Saku Sugawara, and Akiko Aizawa. 2020. Constructing a multi-hop qa dataset for comprehensive evaluation of reasoning steps. arXiv preprint arXiv:2011.01060.
  27. Hongjin Qian, Peitian Zhang, Zheng Liu, Kelong Mao, and Zhicheng Dou. 2024. Memorag: Moving towards next-gen rag via memory-inspired knowledge discovery. arXiv preprint arXiv:2409.05591. — [arXiv:2409.05591](https://arxiv.org/abs/2409.05591) find
  28. Gautier Izacard and Edouard Grave. 2021. Distilling knowledge from reader to retriever for question answering. In ICLR 2021-9th International Conference on Learning Representations. find
  29. Stephen Robertson, Hugo Zaragoza, and 1 others. 2009. The probabilistic relevance framework: Bm25 and beyond. Foundations and Trends® in Information Retrieval, 3(4):333–389. find
  30. Joshua Robinson, Ching-Yao Chuang, Suvrit Sra, and Stefanie Jegelka. 2020. Contrastive learning with hard negative samples. arXiv preprint arXiv:2010.04592. — [arXiv:2010.04592](https://arxiv.org/abs/2010.04592) find
  31. Parth Sarthi, Salman Abdullah, Aditi Tuli, Shubh Khanna, Anna Goldie, and Christopher D. Manning. 2024. RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval. arXiv preprint. ArXiv:2401.18059 [cs].
  32. Yuze Zhao, Jintao Huang, Jinghan Hu, Xingjun Wang, Yunlin Mao, Daoze Zhang, Zeyinzi Jiang, Zhikai Wu, Baole Ai, Ang Wang, Wenmeng Zhou, and Yingda Chen. 2024. Swift:a scalable lightweight infrastructure for fine-tuning. Preprint, arXiv:2408.05517. — [arXiv:2408.05517](https://arxiv.org/abs/2408.05517) find
  33. Jiawei Zhou, Kai Wang, Jianwei Wang, Kunpeng Zhang, and Xuemin Lin. 2025. Comet: An interactive framework for efficient and effective community search via active learning. INFORMS Journal on Computing. find
  34. Weijia Shi, Sewon Min, Michihiro Yasunaga, Minjoon Seo, Rich James, Mike Lewis, Luke Zettlemoyer, and Wen-tau Yih. 2023. Replug: Retrieval-augmented black-box language models. arXiv preprint arXiv:2301.12652.
  35. Saba Sturua, Isabelle Mohr, Mohammad Kalim Akram, Michael Günther, Bo Wang, Markus Krimmel, Feng Wang, Georgios Mastrapas, Andreas Koukounas, Andreas Koukounas, Nan Wang, and Han Xiao. 2024. jina-embeddings-v3: Multilingual embeddings with task lora. Preprint, arXiv:2409.10173.
  36. Jinbo Su, Lingzhe Gao, Wei Li, Shihao Liu, Haojie Lei, Xinyi Wang, Yuanzhao Guo, Ke Wang, Daiting Shi, and Dawei Yin. 2025. Racqc: Advanced retrieval-augmented generation for chinese query correction. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 675–689. find
  37. Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, and 1 others. 2023. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288.
  38. Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, and Ashish Sabharwal. 2022. Musique: Multi-hop questions via single-hop question composition. Transactions of the Association for Computational Linguistics, 10:539–554. find
  39. Zhihao Xu, Rumei Li, Jiahuan Li, Rongxiang Weng, Jingang Wang, Xunliang Cai, and Xiting Wang. 2026. Unlocking implicit experience: Synthesizing tool-use trajectories from text. arXiv preprint arXiv:2601.10355.
  40. Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William W Cohen, Ruslan Salakhutdinov, and Christopher D Manning. 2018. Hotpotqa: A dataset for diverse, explainable multi-hop question answering. arXiv preprint arXiv:1809.09600.
  41. Dun Zhang, Jiacheng Li, Ziyang Zeng, and Fulong Wang. 2025a. Jasper and stella: distillation of sota embedding models. Preprint, arXiv:2412.19048. — [arXiv:2412.19048](https://arxiv.org/abs/2412.19048) find
  42. Yanzhao Zhang, Mingxin Li, Dingkun Long, Xin Zhang, Huan Lin, Baosong Yang, Pengjun Xie, An Yang, Dayiheng Liu, Junyang Lin, Fei Huang, and Jingren Zhou. 2025b. Qwen3 embedding: Advancing text embedding and reranking through foundation models. arXiv preprint arXiv:2506.05176.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zhou-2026-ark-answer-centric-retriever-tuning/