Elamur Research⌕
☰
home / papers / 2026 / SciRAG: Adaptive, Citation-Aware, and Outline-Guided Retrieval and Synthesis for Scientific Literature

SciRAG: Adaptive, Citation-Aware, and Outline-Guided Retrieval and Synthesis for Scientific Literature

Paper · 2026 · Hang Ding, Yilun Zhao, Tiansheng Hu et al.

Summary

Scientific output keeps growing quickly, and existing retrieval-augmented generation (RAG) tools for literature QA have four main weaknesses. They treat citations as flat text, retrieve in a rigid single pass, stitch snippets together without an overall plan, and are often closed and costly. SciRAG is an open-source framework that tackles these problems with three linked components:

  • Adaptive retrieval: a controller chooses between deep sequential exploration and parallel handling of sub-questions.
  • Citation-aware symbolic reasoning: the system follows both forward and backward citation links, groups papers into contribution chains (theory, experiment, method, application), and uses these chains to rerank evidence.
  • Outline-guided synthesis: the system drafts a plan, critiques intermediate answers, fetches missing evidence, and then merges results bottom-up while keeping provenance links.

The authors evaluated SciRAG on open-retrieval benchmarks, including ScholarQA, SciFact, PubMedQA and QASA. It outperformed strong systems such as OpenScholar and PaperQA2 on factual accuracy and synthesis quality, and expert raters preferred its answers. The work matters because it shows that structured use of citation graphs and explicit answer planning make literature synthesis more reliable and verifiable.

Key points

  • The full SciRAG system scores 84.1 on SciFact, 69.0 on ScholarQA-CS and 4.79 on ScholarQA-MULTI.
  • Replacing the symbolic reranker with a dense reranker lowers these scores by 10.7, 10.4 and 0.34. Removing the outline planner lowers them by 8.0, 16.9 and 0.72.
  • Both sequential-only and parallel-only retrieval perform worse than the adaptive mix, with drops of roughly 7–11 points on SciFact and ScholarQA-CS.
  • Retrieval tree depth works best at 1–2 levels. For example, PubMedQA rises from 76.2 at depth 0 to 84.1 at depth 2, and quality falls at depths 3–4 because irrelevant citations add noise.
  • The system is built on general-purpose LLMs (GPT-4o, Llama-3.1) and is released at github.com/yale-nlp/SciRAG. Its stated limitations are computational overhead and a small pool of human annotators.

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 · China149 works · 16,161 citations🎓Yale UniversityNew Haven · United States95 works · 12,765 citations🎓New York University ShanghaiShanghai · China8 works · 523 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksInformation Systems · 1,437 works

🧭 Topics 1Research subjects and languages the work is about.Information retrieval & search · 473 works

Cite

Cite as Hang Ding, Yilun Zhao, Tiansheng Hu, Zihang Wang, Manasi Patwardhan and Arman Cohan. 2026. SciRAG: Adaptive, Citation-Aware, and Outline-Guided Retrieval and Synthesis for Scientific Literature. Proceedings of the 19th Conference of the European Chapter 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.eacl-long.303

@inproceedings{ding2026scirag,
  title = {{SciRAG: Adaptive, Citation-Aware, and Outline-Guided Retrieval and Synthesis for Scientific Literature}},
  author = {Hang Ding, Yilun Zhao, Tiansheng Hu, Zihang Wang, Manasi Patwardhan and Arman Cohan},
  booktitle = {Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.eacl-long.303},
  url = {https://doi.org/10.18653/v1/2026.eacl-long.303}
}

References

25 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Shubham Agarwal, Issam H Laradji, Laurent Charlin, and Christopher Pal. 2024. Litllm: A toolkit for scientific literature review. arXiv preprint arXiv:2402.01788. find
  2. Akari Asai, Jacqueline He, Rulin Shao, Weijia Shi, Amanpreet Singh, Joseph Chee Chang, Kyle Lo, Luca Soldaini, Sergey Feldman, Mike D’arcy, et al. 2024. Openscholar: Synthesizing scientific literature with retrieval-augmented lms. arXiv preprint arXiv:2411.14199. find
  3. Jinheon Baek, Sujay Kumar Jauhar, Silviu Cucerzan, and Sung Ju Hwang. 2025. Researchagent: Iterative research idea generation over scientific literature with large language models.
  4. Iz Beltagy, Kyle Lo, and Arman Cohan. 2019. Scibert: A pretrained language model for scientific text.
  5. Lutz Bornmann and Hans-Dieter Daniel. 2008. What do citation counts measure? a review of studies on citing behavior. Journal of Documentation, 64(1):45–80. find
  6. Lutz Bornmann and Rüdiger Mutz. 2015. Growth rates of modern science: A bibliometric analysis based on the number of publications and cited references. Journal of the association for information science and technology, 66(11):2215–2222. find
  7. Crossref. 2024. Crossref metadata statistics. https://www.crossref.org/. Accessed: 2025-09-29. ↗
  8. Xinyan Guan, Jiali Zeng, Fandong Meng, Chunlei Xin, Yaojie Lu, Hongyu Lin, Xianpei Han, Le Sun, and Jie Zhou. 2025. Deeprag: Thinking to retrieval step by step for large language models. arXiv preprint arXiv:2502.01142. find
  9. Yuntong Hu, Zhihan Lei, Zhongjie Dai, Allen Zhang, Abhinav Angirekula, Zheng Zhang, and Liang Zhao. 2025. Cg-rag: Research question answering by citation graph retrieval-augmented llms. find
  10. Qian Huang, Jian Vora, Percy Liang, and Jure Leskovec. 2023. Mlagentbench: Evaluating language agents on machine learning experimentation. In International Conference on Machine Learning.
  11. Qiao Jin, Bhuwan Dhingra, Zhengping Liu, William Cohen, and Xinghua Lu. 2019. PubMedQA: A dataset for biomedical research question answering. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Proce…
  12. Seungone Kim, Juyoung Suk, Shayne Longpre, Bill Yuchen Lin, Jamin Shin, Sean Welleck, Graham Neubig, Moontae Lee, Kyungjae Lee, and Minjoon Seo. 2024. Prometheus 2: An open source language model specialized in evaluating other language models. In Proceedings of the 2024 Conference on Empirical Metho…
  13. Yoonjoo Lee, Kyungjae Lee, Sunghyun Park, Dasol Hwang, Jaehyeon Kim, Hong-In Lee, and Moontae Lee. 2023. QASA: Advanced question answering on scientific articles. In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Research, pages 19… find
  14. Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, et al. 2020. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in neural information processing systems, 33:9459–9474.
  15. OpenAI, :, Aaron Hurst, Adam Lerer, Adam P. Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, Aleksander M adry, Alex Baker-Whitcomb, Alex Beutel, Alex Borzunov, Alex Carney, Alex Chow, Alex Kirillov, Alex Nichol, Alex Paino, Alex Renzin, Alex… — GPT-4o System Card
  16. Amanpreet Singh, Joseph Chee Chang, Chloe Anastasiades, Dany Haddad, Aakanksha Naik, Amber Tanaka, Angele Zamarron, Cecile Nguyen, Jena D. Hwang, Jason Dunkleberger, Matt Latzke, Smita Rao, Jaron Lochner, Rob Evans, Rodney Kinney, Daniel S. Weld, Doug Downey, and Sergey Feldman. 2025. Ai2 scholar qa… find
  17. Shruti Singh, Nandan Sarkar, and Arman Cohan. 2024. SciDQA: A deep reading comprehension dataset over scientific papers. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 20908–20923, Miami, Florida, USA. Association for Computational Linguistics. find
  18. Michael D. Skarlinski, Sam Cox, Jon M. Laurent, James D. Braza, Michaela Hinks, Michael J. Hammerling, Manvitha Ponnapati, Samuel G. Rodriques, and Andrew D. White. 2024. Language agents achieve superhuman synthesis of scientific knowledge. find
  19. Minyang Tian, Luyu Gao, Shizhuo Dylan Zhang, Xinan Chen, Cunwei Fan, Xuefei Guo, Roland Haas, Pan Ji, Kittithat Krongchon, Yao Li, Shengyan Liu, Di Luo, Yutao Ma, Hao Tong, Kha Trinh, Chenyu Tian, Zihan Wang, Bohao Wu, Yanyu Xiong, Shengzhu Yin, Minhui Zhu, Kilian Lieret, Yanxin Lu, Genglin Liu, Yuf… find
  20. David Wadden, Shanchuan Lin, Kyle Lo, Lucy Lu Wang, Madeleine van Zuylen, Arman Cohan, and Hannaneh Hajishirzi. 2020. Fact or fiction: Verifying scientific claims. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 7534–7550, Online. Association…
  21. Liang Wang, Haonan Chen, Nan Yang, Xiaolong Huang, Zhicheng Dou, and Furu Wei. 2025. Chain-of-retrieval augmented generation. find
  22. Zonglin Yang, Xinya Du, Junxian Li, Jie Zheng, Soujanya Poria, and Erik Cambria. 2024. Large language models for automated open-domain scientific hypotheses discovery. find
  23. Jiasheng Zhang, Jialin Chen, Ali Maatouk, Ngoc Bui, Qianqian Xie, Leandros Tassiulas, Jie Shao, Hua Xu, and Rex Ying. 2024. Litfm: A retrieval augmented structure-aware foundation model for citation graphs. arXiv preprint arXiv:2409.12177. — [arXiv:2409.12177](https://arxiv.org/abs/2409.12177) find
  24. Yuxiang Zheng, Shichao Sun, Lin Qiu, Dongyu Ru, Cheng Jiayang, Xuefeng Li, Jifan Lin, Binjie Wang, Yun Luo, Renjie Pan, et al. 2024. Openresearcher: Unleashing ai for accelerated scientific research. arXiv preprint arXiv:2408.06941. find
  25. Ziyuan Zhuang, Zhiyang Zhang, Sitao Cheng, Fangkai Yang, Jia Liu, Shujian Huang, Qingwei Lin, Saravan Rajmohan, Dongmei Zhang, and Qi Zhang. 2024. Efficientrag: Efficient retriever for multi-hop question answering. arXiv preprint arXiv:2408.04259. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/ding-2026-scirag-adaptive-citation-aware-outline/