Elamur Research⌕
☰
home / papers / 2026 / Reasoning with Ontology Graph: Toward Type-Constrained Knowledge Graph Question Answering

Reasoning with Ontology Graph: Toward Type-Constrained Knowledge Graph Question Answering

Paper · 2026 · Yongxue Shan, Jie Peng, Zixuan Dong et al.

Summary

The paper tackles multi-hop knowledge graph question answering (KGQA) with LLMs. It targets two problems in earlier work. LLM-built ontologies label the same entity at uneven levels of abstraction. Subgraph-based search also lets in many paths ending in the wrong entity type, so the number of candidate paths explodes. The authors build OntGQA on a reusable, relation-centric ontology graph taken from the Freebase schema rather than generated by a model. In this graph, every relation carries canonical head and tail types. Reasoning runs in three stages:

  1. A fine-tuned planner predicts likely head–tail type pairs for the question.
  2. These type pairs restrict hop-by-hop path retrieval in the KG.
  3. A contrastively trained judge accepts or rejects each candidate answer together with its path. A generator answers only when the judge rejects every candidate.

Fixing both ends of the reasoning in type space shrinks the search and produces chains grounded in the ontology. OntGQA reports state-of-the-art Hit@1 on WebQSP and CWQ. It also trains on fewer samples and runs inference faster than the GCR baseline, which matters for building more faithful and efficient KGQA systems.

Key points

  • Ontology graph construction:
    • It is parsed from the Freebase RDF dump, with type.property.schema giving the head type and type.property.expected_type giving the tail type.
    • Administrative types such as common.topic are removed.
    • A few relations with missing types are filled in using types inferred from training-split entities only.
  • Retrieval: it covers 1-hop paths and 2-hop paths with type "boundary gates," and backs off to 3–4 hops when nothing is found.
  • Hit@1 gains: WebQSP rises from 87.7% to 91.5%, while CWQ climbs from 67.6% to 74.6%.
  • Backbones: Qwen2-7B gave the best overall F1. GPT-4o and DeepSeek-v3 matched it on Hit@1 but had lower F1.
  • Efficiency compared with GCR:
    • It used 19,692 training samples on WebQSP and 100,693 on CWQ, roughly 30–45% fewer.
    • Training took 7.9 GPU hours.
    • Average inference time was 2.57 s.
  • Limitations: the method has been tested only on Freebase.

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

🎓National University of Defense TechnologyChangsha · China10 works · 931 citations

 Fields, topics and concepts

📚 Shelves 4The 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 works

🧭 Topics 3Research subjects and languages the work is about.Reasoning · 1,426 worksQuestion answering · 441 worksKnowledge graphs & ontologies · 192 works

Cite

Cite as Yongxue Shan, Jie Peng, Zixuan Dong, Fei Hu and Xiaodong Wang. 2026. Reasoning with Ontology Graph: Toward Type-Constrained Knowledge Graph 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. DOI 10.18653/v1/2026.acl-long.489

@inproceedings{shan2026reasoning,
  title = {{Reasoning with Ontology Graph: Toward Type-Constrained Knowledge Graph Question Answering}},
  author = {Yongxue Shan, Jie Peng, Zixuan Dong, Fei Hu and Xiaodong 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.489},
  url = {https://doi.org/10.18653/v1/2026.acl-long.489}
}

References

40 references, 10 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Hamed Babaei Giglou, Jennifer D’Souza, and Sören Auer. 2023. LLMs4OL: Large language models for ontology learning. In International Semantic Web Conference, pages 408–427. Springer. find
  2. Kurt Bollacker, Colin Evans, Praveen Paritosh, Tim Sturge, and Jamie Taylor. 2008. Freebase: a collaboratively created graph database for structuring human knowledge. In Proceedings of the 2008 ACM SIG-MOD international conference on Management of data, pages 1247–1250. base question answering by le… find
  3. Jinhao Jiang, Kun Zhou, Xin Zhao, and Ji-Rong Wen. 2023. Unikgqa: Unified retrieval and reasoning for solving multi-hop question answering over knowledge graph. In The Eleventh International Conference on Learning Representations. find
  4. Deyan Chen and Hong Zhao. 2018. Research on the method of extracting domain knowledge from the freebase RDF dumps. IEEE Access, 6:50306–50322. find
  5. Longquan Jiang, Junbo Huang, Cedric Moller, and Ricardo Usbeck. 2025. Ontology-guided, hybrid prompt learning for generalization in knowledge graph question answering. In 2025 19th International Conference on Semantic Computing (ICSC), pages 28–35. IEEE Computer Society. find
  6. Yixuan Chen, Dongsheng Li, Peng Zhang, Jie Sui, Qin Lv, Lu Tun, and Li Shang. 2022. Cross-modal ambiguity learning for multimodal fake news detection. In Proceedings of the ACM on Web Conference 2022, pages 2897–2905. find
  7. Vamsi Krishna Kommineni, Birgitta König-Ries, and Sheeba Samuel. 2024. From human experts to machines: An LLM supported approach to ontology and knowledge graph construction. arXiv preprint arXiv:2403.08345. — [arXiv:2403.08345](https://arxiv.org/abs/2403.08345) find
  8. Zhongwu Chen, Long Bai, Zixuan Li, Zhen Huang, Xiaolong Jin, and Yong Dou. 2024. A new pipeline for knowledge graph reasoning enhanced by large language models without fine-tuning. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 1366–1381. find
  9. Ke Liang, Lingyuan Meng, Yue Liu, Meng Liu, Wei Wei, Suyuan Liu, Wenxuan Tu, Siwei Wang, Sihang Zhou, and Xinwang Liu. 2024. Simple yet effective: Structure guided pre-trained transformer for multimodal knowledge graph reasoning. In Proceedings of the 32nd ACM International Conference on Multimedia,… find
  10. Roi Cohen, Mor Geva, Jonathan Berant, and Amir Globerson. 2023. Crawling the internal knowledge-base of language models. In Findings of the Association for Computational Linguistics: EACL 2023, pages 1856–1869. Association for Computational Linguistics.
  11. DeepSeek-AI, Aixin Liu, Bei Feng, Bing Xue, Bing-Li Wang, Bochao Wu, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, Damai Dai, Daya Guo, Dejian Yang, Deli Chen, Dong-Li Ji, Erhang Li, Fangyun Lin, Fucong Dai, and 179 others. 2024. DeepSeek-V3 Technical Report. ArXiv, abs/2412.19… find
  12. Giacomo Frisoni, Alessio Cocchieri, Alex Presepi, Gianluca Moro, and Zaiqiao Meng. 2024. To generate or to retrieve? on the effectiveness of artificial contexts for medical open-domain question answering. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Vol… find
  13. Maurice Funk, Simon Hosemann, Jean Christoph Jung, and Carsten Lutz. 2023. Towards ontology construction with language models. In Joint proceedings of the 1st workshop on Knowledge Base Construction from Pre-Trained Language Models (KBC-LM) and the 2nd challenge on Language Models for Knowledge Base…
  14. Gaole He, Yunshi Lan, Jing Jiang, Wayne Xin Zhao, and Ji-Rong Wen. 2021. Improving multi-hop knowledge Ben Liu, Jihai Zhang, Fangquan Lin, Cheng Yang, Min Peng, and Wotao Yin. 2025a. Symagent: A neural-symbolic self-learning agent framework for complex reasoning over knowledge graphs. In Proceedings… find
  15. Runxuan Liu, Luobei Luobei, Jiaqi Li, Baoxin Wang, Ming Liu, Dayong Wu, Shijin Wang, and Bing Qin. 2025b. Ontology-guided reverse thinking makes large language models stronger on knowledge graph question answering. In Proceedings of the 63rd Annual Meeting of the Association for Computational Lingui…
  16. Zhiqiang Liu, Chengtao Gan, Junjie Wang, Yichi Zhang, Zhongpu Bo, Mengshu Sun, Huajun Chen, and Wen Zhang. 2025c. Ontotune: Ontology-driven self-training for aligning large language models. In Proceedings of the ACM on Web Conference 2025, pages 119–133. find
  17. Linhao Luo, Yuanfang Li, Gholamreza Haffari, and Shirui Pan. 2024. Reasoning on graphs: Faithful and interpretable large language model reasoning. In The Twelfth International Conference on Learning Representations. find
  18. Linhao Luo, Zicheng Zhao, Gholamreza Haffari, Yuan-Fang Li, Chen Gong, and Shirui Pan. 2025. Graph-constrained reasoning: Faithful reasoning on knowledge graphs with large language models. In Forty-second International Conference on Machine Learning. find
  19. Costas Mavromatis and George Karypis. 2025. GNN-RAG: Graph neural retrieval for efficient large language model reasoning on knowledge graphs. In Findings of the Association for Computational Linguistics: ACL 2025, pages 16682–16699. find
  20. Duncan, Petra Fey, Pascale Gaudet, Nomi L. Harris, marcin p. joachimiak, Leila Kiani, Tiago Lubiana, Monica C. Munoz-Torres, Shawn T. O’Neil, David Osumi-Sutherland, Aleix Puig, and 11 others. 2024. Dynamic retrieval augmented generation of ontologies using artificial intelligence (dragon-ai). Journ… find
  21. Meta. 2024. Build the Future of AI with Meta Llama 3. find
  22. Microsoft. 2015. Fastrdfstore-data. find
  23. Alexander Miller, Adam Fisch, Jesse Dodge, Amir-Hossein Karimi, Antoine Bordes, and Jason Weston. 2016. Key-value memory networks for directly reading documents. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 1400–1409. find
  24. George A Miller. 1995. WordNet: a lexical database for English. Communications of the ACM, 38(11):39– 41.
  25. OpenAI. 2022. Introducing ChatGPT. find
  26. OpenAI. 2024. Hello gpt-4o. find
  27. Shirui Pan, Linhao Luo, Yufei Wang, Chen Chen, Jiapu Wang, and Xindong Wu. 2024. Unifying large language models and knowledge graphs: A roadmap. IEEE Transactions on Knowledge and Data Engineering, 36(7):3580–3599.
  28. Haitian Sun, Bhuwan Dhingra, Manzil Zaheer, Kathryn Mazaitis, Ruslan Salakhutdinov, and William Cohen. 2018. Open domain question answering using early fusion of knowledge bases and text. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 4231–4242. find
  29. Jiashuo Sun, Chengjin Xu, Lumingyuan Tang, Saizhuo Wang, Chen Lin, Yeyun Gong, Lionel Ni, Heung-Yeung Shum, and Jian Guo. 2024. Think-on-graph: Deep and responsible reasoning of large language model on knowledge graph. In The Twelfth International Conference on Learning Representations. find
  30. Alon Talmor and Jonathan Berant. 2018. The web as a knowledge-base for answering complex questions. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers), pages 641–651.
  31. Xingyu Tan, Xiaoyang Wang, Qing Liu, Xiwei Xu, Xin Yuan, and Wenjie Zhang. 2025. Paths-over-graph: Knowledge graph empowered large language model reasoning. In Proceedings of the ACM on Web Conference 2025, pages 3505–3522.
  32. 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
  33. Jimmy Donal Wales and Larry Sanger. 2001. Wikipedia. find
  34. Yilin Wen, Zifeng Wang, and Jimeng Sun. 2024. Mindmap: Knowledge graph prompting sparks graph of thoughts in large language models. In 62nd Annual Meeting of the Association for Computational Linguistics, ACL 2024, pages 10370–10388. Association for Computational Linguistics (ACL).
  35. Wen-tau Yih, Matthew Richardson, Christopher Meek, Ming-Wei Chang, and Jina Suh. 2016. The value of semantic parse labeling for knowledge base question answering. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 201–206.
  36. Donghan Yu, Sheng Zhang, Patrick Ng, Henghui Zhu, Alexander Hanbo Li, Jun Wang, Yiqun Hu, William Yang Wang, Zhiguo Wang, and Bing Xiang. 2022. Decaf: Joint decoding of answers and logical forms for question answering over knowledge bases. In The Eleventh International Conference on Learning Represe… find
  37. Antonio Zaitoun, Tomer Sagi, Szymon Wilk, and Mor Peleg. 2023. Can large language models augment a biomedical ontology with missing concepts and relations? arXiv preprint arXiv:2311.06858. — [arXiv:2311.06858](https://arxiv.org/abs/2311.06858) find
  38. Jing Zhang, Xiaokang Zhang, Jifan Yu, Jian Tang, Jie Tang, Cuiping Li, and Hong Chen. 2022. Subgraph retrieval enhanced model for multi-hop knowledge base question answering. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 577… find
  39. WANG Ting, WANG Na, CUI Yunpeng, and LIU Juan. 2023. Agricultural technology knowledge intelligent question-answering system based on large language model. Smart agriculture, 5(4):105. find
  40. Sabrina Toro, Anna V. Anagnostopoulos, Sue Bello, Kai Blumberg, Rhiannon Cameron, Leigh Carmody, Alexander D. Diehl, Damion M. Dooley, William find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/shan-2026-reasoning-ontology-graph-type-constrained/