Elamur Research⌕
☰
home / papers / 2026 / SemEval-2026 Task 12: Knowledge Graph with hyperbolic embedding in Abductive Event Reasoning

SemEval-2026 Task 12: Knowledge Graph with hyperbolic embedding in Abductive Event Reasoning

Paper · 2026 · Mingkai Wang, Varun Ojha and Huizhi Liang

Summary

The paper describes a system for SemEval-2026 Task 12, Abductive Event Reasoning (AER). Given a real-world news event and a set of retrieved documents, some of them distractors, a model must pick the most plausible direct cause from four candidate options (A–D). More than one option can be correct. The authors compare four single-modality baselines:

  • a shallow classifier trained on frozen embeddings;
  • a fine-tuned DeBERTa-v3 multiple-choice model, alongside zero-shot UnifiedQA;
  • a causal knowledge graph expanded with COMET-ATOMIC 2020 hypotheses, embedded with Euclidean RotatE, with retrieved triples turned into text for DeepSeek-chat;
  • the same graph embedded in Lorentz-model hyperbolic space with RotH.

On the development set, DeBERTa-v3 clearly performs best. COMET-based graph expansion substantially narrows the gap between the knowledge-graph systems and DeBERTa-v3. Hyperbolic geometry adds almost nothing on top of the Euclidean version. The authors conclude that well-verbalized causal knowledge and strong discriminative MCQA backbones matter more than elaborate graph geometry. This is a useful finding for anyone weighing KG-augmented LLM pipelines for causal reasoning.

Key points

  • Data: the official splits are sample (200 instances), train (1,819), dev (400) and test (612). Topics include politics, finance, technology and emergencies. Scoring gives 1 point for an exact match, 0.5 for a partial match and 0 for a wrong answer.
  • Dev results:
    • DeBERTa-v3 base: 0.7250 (49.5% exact match, 4.5% error rate)
    • Baseline 1: 0.3925
    • KG + claude-3-haiku without COMET: 0.4625
    • KG + DeepSeek-chat + COMET (RotatE): 0.6088
    • Hyperbolic RotH KG: 0.6100
  • Causal relations: the knowledge graph uses COMET-derived relations (Causes, isBefore, isAfter, xEffect, oEffect) instead of generic factual predicates.
  • Ablation: knowledge-graph embedding with retrieval only, without COMET-based prompts, scored far below the prompt-fusion variants.
  • Tools and code: the system is built with PyTorch, Transformers and Geoopt, and the code is published at github.com/MingkaiW/SemEval2026-task12.

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

🎓Newcastle UniversityNewcastle upon Tyne · United Kingdom18 works · 619 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 worksComputer Vision and Pattern Recognition · 2,869 works

🧭 Topics 2Research subjects and languages the work is about.Reasoning · 1,426 worksKnowledge graphs & ontologies · 192 works

Cite

Cite as Mingkai Wang, Varun Ojha and Huizhi Liang. 2026. SemEval-2026 Task 12: Knowledge Graph with hyperbolic embedding in Abductive Event Reasoning. Proceedings of the 20th International Workshop on Semantic Evaluation (2026).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.semeval-1.114

@inproceedings{wang2026semeval,
  title = {{SemEval-2026 Task 12: Knowledge Graph with hyperbolic embedding in Abductive Event Reasoning}},
  author = {Mingkai Wang, Varun Ojha and Huizhi Liang},
  booktitle = {Proceedings of the 20th International Workshop on Semantic Evaluation (2026)},
  year = {2026},
  doi = {10.18653/v1/2026.semeval-1.114},
  url = {https://doi.org/10.18653/v1/2026.semeval-1.114}
}

References

10 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Haoliang Wang, Tong Yu, Julian McAuley, Wei Ai, and Furong Huang. 2025b. Large language models and causal inference in collaboration: A comprehensive survey. In Proceedings of the 2025 Conference of the North American Chapter of the Association for Computational Linguistics (Volume 1: Long Papers),… find
  2. Martina Miliani, Serena Auriemma, Alessandro Bondielli, Emmanuele Chersoni, Lucia Passaro, Irene Sucameli, and Alessandro Lenci. 2025. Explica: Evaluating explicit causal reasoning in large language models. In Findings of the Association for Computational Linguistics: ACL 2025, pages 17335–17355, Vi… find
  3. Zequn Sun, Muhao Chen, Wei Hu, Chengming Wang, Jian Dai, and Wei Zhang. 2020. Knowledge association with hyperbolic knowledge graph embeddings. In Proceedings of the 2020 conference on empirical methods in natural language processing (EMNLP), pages 5704–5716. find
  4. Chandra Bhagavatula, Ronan Le Bras, Chaitanya Malaviya, Keisuke Sakaguchi, Ari Holtzman, Hannah Rashkin, Doug Downey, Wen-tau Yih, and Yejin Choi. 2020. Abductive commonsense reasoning. In International Conference on Learning Representations.
  5. Pengfei Cao, Yubo Chen, Mingxuan Yang, Chenlong Zhang, Mingxuan Liu, Kang Liu, and Jun Zhao. 2026. Semeval-2026 task 12: Abductive event reasoning: Towards real-world event causal inference for large language models. find
  6. Ines Chami, Adva Wolf, Da-Cheng Juan, Frederic Sala, Sujith Ravi, and Christopher Ré. 2020. Low-dimensional hyperbolic knowledge graph embeddings. In Proceedings of the 58th annual meeting of the association for computational linguistics, pages 6901–6914.
  7. Haoang Chi, He Li, Wenjing Yang, Feng Liu, Long Lan, Xiaoguang Ren, Tongliang Liu, and Bo Han. 2024. Unveiling causal reasoning in large language models: Reality or mirage? In Advances in Neural Information Processing Systems. find
  8. Emre Kiciman, Robert Osazuwa Ness, Amit Sharma, and Chenhao Tan. 2024. Causal reasoning and large language models: Opening a new frontier for causality. Transactions on Machine Learning Research. find
  9. Junming Liu, Siyuan Meng, Yanting Gao, Song Mao, Pinlong Cai, Guohang Yan, Yirong Chen, Zilin Bian, Ding Wang, and Botian Shi. 2025a. Aligning vision to language: Annotation-free multimodal knowledge graph construction for enhanced llms reasoning. In Proceedings of the IEEE/CVF International Confere… find
  10. Xiaoyu Liu, Paiheng Xu, Junda Wu, Jiaxin Yuan, Yifan Yang, Yuhang Zhou, Fuxiao Liu, Tianrui Guan, find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/wang-2026-semeval-2026-task-12-knowledge/