Elamur Research⌕
☰
home / papers / 2026 / Learning to Think on Hypergraph: HyperCoT for Structure-Guided N-ary Knowledge Graph Completion

Learning to Think on Hypergraph: HyperCoT for Structure-Guided N-ary Knowledge Graph Completion

Paper · 2026 · Mengxue Yang, Jinming Li, Chun Yang et al.

Summary

The paper addresses n-ary knowledge graph completion. The task is to predict a missing entity in a fact that links several entities, each filling a distinct semantic role. Such a fact is a hyperedge in a knowledge hypergraph. The authors identify two gaps. Embedding and GNN hypergraph models score hyperedges locally and discard intermediate propagation states. LLM-based methods work well for triple-based KGC, but here they see only the query hyperedge or one pooled signal, which yields fluent but structurally wrong answers.

Their method, HyperCoT, works in three steps:

  • It builds role- and relation-aware hyperedge states using low-rank projections and attention over role-pair interactions.
  • It reasons across several hops and keeps the state at each depth.
  • It merges the strongest paths into a Graphical Chain-of-Thought (Graph-CoT) and injects this path-level evidence into the tokens of the query and candidate entities. An LLM then ranks the candidates.

On JF17K, WikiPeople and FB-AUTO, HyperCoT outperforms strong n-ary KGC baselines. Gains are largest for high-arity facts and sparsely connected entities. It also produces interpretable multi-hop reasoning traces, showing that structural evidence must be fed to LLMs explicitly for hypergraph link prediction.

Key points

  • Benchmarks are JF17K, WikiPeople and FB-AUTO. The full model reaches MRR 0.599, 0.451 and 0.904, respectively, with H@10 of 0.743, 0.612 and 0.927.
  • Ablations show the largest drop on all three datasets when progressive multi-hop reasoning is removed (e.g., JF17K MRR falls to 0.564). Removing only the Graph-CoT representation still lowers scores, so the gains do not come from textual structural prompting alone.
  • On the 20% lowest-degree entities, HyperCoT beats HyConvE and HyCubE+ on all datasets. On JF17K its MRR is 0.482 versus 0.458 for HyCubE+.
  • Robustness was tested by randomly dropping 10–40% of hyperedges. On JF17K, HyperCoT's MRR degrades less than that of the baselines.
  • Default settings are a reasoning depth of L = 3 and K = 5 Graph-CoT paths. Larger values bring no gain or add noise.

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

🏛️Chinese Academy of SciencesBeijing · China246 works · 24,952 citations🎓University of Chinese Academy of SciencesBeijing · China188 works · 8,744 citations🔬Institute of SoftwareBeijing · China27 works · 4,980 citations

 Fields, topics and concepts

📚 Shelves 4The 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 works

🧭 Topics 1Research subjects and languages the work is about.Knowledge graphs & ontologies · 211 works

Cite

Cite as Mengxue Yang, Jinming Li, Chun Yang, Jiaqi Zhu, Jiafan Li, Guanhua Zhang and Ying Li. 2026. Learning to Think on Hypergraph: HyperCoT for Structure-Guided N-ary Knowledge Graph Completion. 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.1172

@inproceedings{yang2026learning,
  title = {{Learning to Think on Hypergraph: HyperCoT for Structure-Guided N-ary Knowledge Graph Completion}},
  author = {Mengxue Yang, Jinming Li, Chun Yang, Jiaqi Zhu, Jiafan Li, Guanhua Zhang and Ying Li},
  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.1172},
  url = {https://doi.org/10.18653/v1/2026.acl-long.1172}
}

References

36 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Antoine Bordes, Nicolas Usunier, Alberto García-Durán, Jason Weston, and Oksana Yakhnenko. 2013. Translating embeddings for modeling multi-relational data. In Advances in Neural Information Processing Systems 26: 27th Annual Conference on Neural Information Processing Systems 2013. Proceedings of a…
  2. Zhe Chen, Yuehan Wang, Bin Zhao, Jing Cheng, Xin Zhao, and Zongtao Duan. 2020. Knowledge graph completion: A review. IEEE Access, 8:192435– 192456. find
  3. Zirui Chen, Xin Wang, Chenxu Wang, and Jianxin Li. 2022. Explainable link prediction in knowledge hypergraphs. In Proceedings of the 31st ACM International Conference on Information & Knowledge Management, Atlanta, GA, USA, October 17-21, 2022, pages 262–271. ACM. find
  4. Zirui Chen, Xin Wang, Chenxu Wang, and Zhao Li. 2023. Poskhg: A position-aware knowledge hypergraph model for link prediction. Data Sci. Eng., 8(2):135–145. find
  5. Janneth Chicaiza and Priscila Valdiviezo Díaz. 2021. A comprehensive survey of knowledge graph-based recommender systems: Technologies, development, and contributions. Inf., 12(6):232. find
  6. Bahare Fatemi, Perouz Taslakian, David Vázquez, and David Poole. 2020. Knowledge hypergraphs: Prediction beyond binary relations. In Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence, IJCAI 2020, pages 2191–2197. ijcai.org. find
  7. Bahare Fatemi, Perouz Taslakian, David Vázquez, and David Poole. 2023. Knowledge hypergraph embedding meets relational algebra. J. Mach. Learn. Res., 24:105:1–105:34. find
  8. Yifan Feng, Chengwu Yang, Xingliang Hou, Shaoyi Du, Shihui Ying, Zongze Wu, and Yue Gao. 2024. Beyond graphs: Can large language models comprehend hypergraphs? Preprint, arXiv:2410.10083. — [arXiv:2410.10083](https://arxiv.org/abs/2410.10083) find
  9. Google DeepMind. 2024. Gemma 2 technical report. Technical report, Google DeepMind. find
  10. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Arthur Hinsvark, and 542 o… — The Llama 3 Herd of Models
  11. Saiping Guan, Xiaolong Jin, Jiafeng Guo, Yuanzhuo Wang, and Xueqi Cheng. 2023. Link prediction on n-ary relational data based on relatedness evaluation. IEEE Trans. Knowl. Data Eng., 35(1):672–685. find
  12. Saiping Guan, Xiaolong Jin, Yuanzhuo Wang, and Xueqi Cheng. 2019. Link prediction on n-ary relational data. In The World Wide Web Conference, WWW ’19, page 583–593, New York, NY, USA. Association for Computing Machinery. find
  13. Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2022. Lora: Low-rank adaptation of large language models. In The Tenth International Conference on Learning Representations, ICLR 2022, Virtual Event, April 25-29, 2022. OpenReview.net.
  14. Albert Q. Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, Lélio Renard Lavaud, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix,… find
  15. Zhao Li, Xin Wang, Jun Zhao, Wenbin Guo, and Jianxin Li. 2025. Hycube: Efficient knowledge hypergraph 3d circular convolutional embedding. IEEE Trans. Knowl. Data Eng., 37(4):1902–1914. find
  16. Yang Liu, Xiaobin Tian, Zequn Sun, and Wei Hu. 2024. Finetuning generative large language models with discrimination instructions for knowledge graph completion. In The Semantic Web - ISWC 2024 - 23rd International Semantic Web Conference, Baltimore, MD, USA, November 11-15, 2024, Proceedings, Part… find
  17. Yu Liu, Quanming Yao, and Yong Li. 2021. Role-aware modeling for n-ary relational knowledge bases. In Proceedings of the Web Conference 2021, WWW ’21, page 2660–2671, New York, NY, USA. Association for Computing Machinery. find
  18. Haoran Luo, Haihong E, Guanting Chen, Yandan Zheng, Xiaobao Wu, Yikai Guo, Qika Lin, Yu Feng, Ze-min Kuang, Meina Song, Yifan Zhu, and Luu Anh Tuan. 2025. Hypergraphrag: Retrieval-augmented generation with hypergraph-structured knowledge representation. CoRR, abs/2503.21322. — [arXiv:2503.21322](https://arxiv.org/abs/2503.21322) find
  19. Vinod Nair and Geoffrey E. Hinton. 2010. Rectified linear units improve restricted boltzmann machines. In Proceedings of the 27th International Conference on Machine Learning (ICML-10), June 21-24, 2010, Haifa, Israel, pages 807–814. Omnipress. find
  20. OpenAI. 2024. GPT-4o mini: Advancing cost-efficient intelligence. find
  21. Apoorv Saxena, Adrian Kochsiek, and Rainer Gemulla. 2022. Sequence-to-sequence knowledge graph completion and question answering. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), ACL 2022, Dublin, Ireland, May 22-27, 2022, pages 2814… find
  22. Zhiqing Sun, Zhi-Hong Deng, Jian-Yun Nie, and Jian Tang. 2019. Rotate: Knowledge graph embedding by relational rotation in complex space. In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019. OpenReview.net.
  23. Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton-Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, and 49 others… — Llama 2: Open Foundation and Fine-Tuned Chat Models
  24. Chenxu Wang, Xin Wang, Zhao Li, Zirui Chen, and Jianxin Li. 2023a. Hyconve: A novel embedding model for knowledge hypergraph link prediction with convolutional neural networks. In Proceedings of the ACM Web Conference 2023, WWW 2023, Austin, TX, USA, 30 April 2023 - 4 May 2023, pages 188–198. ACM. find
  25. Yu Wang, Feng Ye, Binquan Li, Gaoyang Jin, Dong Xu, and Fengsheng Li. 2023b. Urbanfloodkg: An urban flood knowledge graph system for risk assessment. In Proceedings of the 32nd ACM International Conference on Information and Knowledge Management, CIKM ’23, page 2574–2584, New York, NY, USA. Associat… find
  26. Yanbin Wei, Qiushi Huang, James T. Kwok, and Yu Zhang. 2024. KICGPT: large language model with knowledge in context for knowledge graph completion. CoRR, abs/2402.02389. find
  27. Jianfeng Wen, Jianxin Li, Yongyi Mao, Shini Chen, and Richong Zhang. 2016. On the representation and embedding of knowledge bases beyond binary relations. In Proceedings of the Twenty-Fifth International Joint Conference on Artificial Intelligence, IJCAI’16, page 1300–1307. AAAI Press. find
  28. Xin Xie, Ningyu Zhang, Zhoubo Li, Shumin Deng, Hui Chen, Feiyu Xiong, Mosha Chen, and Huajun Chen. 2022. From discrimination to generation: Knowledge graph completion with generative transformer. In Companion of The Web Conference 2022, Virtual Event / Lyon, France, April 25 - 29, 2022, pages 162–16… find
  29. An Yang, Baosong Yang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Zhou, Chengpeng Li, Chengyuan Li, Dayiheng Liu, Fei Huang, Guanting Dong, Haoran Wei, Huan Lin, Jialong Tang, Jialin Wang, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Ma, and 43 others. 2024. Qwen2 technical report. Preprint, arXiv:240… find
  30. Bishan Yang, Wen-tau Yih, Xiaodong He, Jianfeng Gao, and Li Deng. 2015. Embedding entities and relations for learning and inference in knowledge bases. In 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings.
  31. Mengxue Yang, Chun Yang, Jiaqi Zhu, Jiafan Li, Jingqi Zhang, Yuyang Li, and Ying Li. 2025a. SLiNT: Structure-aware language model with injection and contrastive training for knowledge graph completion. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 13658–13671, Suzho… find
  32. Rui Yang, Jiahao Zhu, Jianping Man, Hongze Liu, Li Fang, and Yi Zhou. 2025b. GS-KGC: A generative subgraph-based framework for knowledge graph completion with large language models. Inf. Fusion, 117:102868. find
  33. Mohammad Yani and Adila Alfa Krisnadhi. 2021. Challenges, techniques, and trends of simple knowledge graph question answering: A survey. Inf., 12(7):271. find
  34. Fuxiang Zhang, Xin Wang, Zhao Li, and Jianxin Li. 2020. Transrhs: A representation learning method for knowledge graphs with relation hierarchical structure. In Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence, IJ-CAI 2020, pages 2987–2993. ijcai.org. find
  35. Richong Zhang, Junpeng Li, Jiajie Mei, and Yongyi Mao. 2018. Scalable instance reconstruction in knowledge bases via relatedness affiliated embedding. In Proceedings of the 2018 World Wide Web Conference on World Wide Web, WWW 2018, Lyon, France, April 23-27, 2018, pages 1185–1194. ACM. find
  36. Xue Zhou, Bei Hui, Ilana Zeira, Hao Wu, and Ling Tian. 2023. Dynamic relation learning for link prediction in knowledge hypergraphs. Appl. Intell., 53(22):26580–26591. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yang-2026-learning-think-hypergraph-hypercot-structure/