Elamur Research⌕
☰
home / papers / 2026 / Emotion–Cause Pair Extraction in Conversations via Semantic Decoupling and Alignment

Emotion–Cause Pair Extraction in Conversations via Semantic Decoupling and Alignment

Paper · 2026 · Tianxiang Ma, Weijie Feng, Xinyu Wang et al.

Summary

The paper addresses Emotion–Cause Pair Extraction in Conversations (ECPEC): finding which utterances in a dialogue trigger which emotional utterances. Earlier systems either select candidates first and pair them afterwards, or classify concatenated utterance embeddings pair by pair. The authors argue that both designs share two weaknesses. They put emotion and cause information into one representation space. They also judge each pair in isolation, which misses the fact that one emotion can have several causes and one cause can feed several emotions.

Their framework, SCALE, starts from RoBERTa utterance embeddings combined with speaker embeddings. It builds a single conversation graph and runs two separate attention-based graph encoders over it, one producing emotion-oriented representations and the other cause-oriented ones. Pair extraction is then cast as a global matching between these two views, which is solved with optimal transport. Emotion extraction and cause extraction act as auxiliary training signals.

On three benchmarks, SCALE reports state-of-the-art results and also beats few-shot prompted LLMs. This suggests that explicitly modelling dialogue structure still pays off for conversational causality.

Key points

  • Graph construction: the conversation graph has three edge types, each with its own initial weights.
    • Global contextual edges are added when cosine similarity passes a threshold.
    • Local temporal edges connect utterances inside a sliding window and decay exponentially with distance.
    • Intra-speaker edges link utterances by the same speaker.
    • The adjacency matrix is learnable.
  • Benchmarks: evaluation uses RECCON-DD, RECCON-IE and ECF. Code is published at github.com/CoCoSphere/SCALE.
  • Comparison with LLMs: the LLMs were prompted with four in-context examples and JSON-constrained output. SCALE scores the highest F1 on both datasets tested.
ModelRECCON-DD F1ECF F1
SCALE58.8357.70
Gemini-3-pro-preview56.0855.42
GPT-5.1 Instant55.2654.76
DeepSeek-V3.247.1142.81
  • Error analysis:
    • On ECF, recall drops as the number of causes per emotion grows.
    • Causes far from the emotion in the dialogue remain hard to capture.
    • t-SNE plots show that SCALE separates emotion and cause features, whereas MECPE-2steps mixes them.
  • Limitations: the method is text-only and ignores acoustic and visual signals. The authors also warn that annotation bias in the training data may carry over into the model.

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

🎓Hefei University of TechnologyHefei · China10 works · 1,615 citations

 Fields, topics and concepts

📚 Shelves 6The 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 worksPsychology · 1,552 worksExperimental and Cognitive Psychology · 870 works

🧭 Topics 1Research subjects and languages the work is about.Sentiment & emotion · 972 works

Cite

Cite as Tianxiang Ma, Weijie Feng, Xinyu Wang and Zhiyong Cheng. 2026. Emotion–Cause Pair Extraction in Conversations via Semantic Decoupling and Alignment. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.617

@inproceedings{ma2026emotion,
  title = {{Emotion–Cause Pair Extraction in Conversations via Semantic Decoupling and Alignment}},
  author = {Tianxiang Ma, Weijie Feng, Xinyu Wang and Zhiyong Cheng},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.617},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.617}
}

References

49 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Alexander Pak and Patrick Paroubek. 2010. Twitter as a corpus for sentiment analysis and opinion mining. In LREC, Valletta, Malta. European Language Resources Association (ELRA).
  2. Jiaming An, Zixiang Ding, Ke Li, and Rui Xia. 2023. Global-view and speaker-aware emotion cause extraction in conversations. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 31:3814– 3823. find
  3. Carlos Busso, Murtaza Bulut, Chi-Chun Lee, Abe Kazemzadeh, Emily Mower, Samuel Kim, Jeannette N. Chang, Sungbok Lee, and Shrikanth S. Narayanan. 2008. Iemocap: Interactive emotional dyadic motion capture database. Language Resources and Evaluation, 42(4):335–359. find
  4. Erik Cambria, Soujanya Poria, Devamanyu Hazarika, and Kenneth Kwok. 2018. Senticnet 5: Discovering conceptual primitives for sentiment analysis by means of context embeddings. Proceedings of the AAAI Conference on Artificial Intelligence, 32(1). find
  5. Shunjie Chen, Xiaochuan Shi, Jingye Li, Shengqiong Wu, Hao Fei, Fei Li, and Donghong Ji. 2022. Joint alignment of multi-task feature and label spaces for emotion cause pair extraction. In Proceedings of the 29th International Conference on Computational Linguistics, pages 6955–6965, Gyeongju, Republ… find
  6. Ying Chen, Wenjun Hou, Shoushan Li, Caicong Wu, and Xiaoqiang Zhang. 2020. End-to-end emotion-cause pair extraction with graph convolutional network. In Proceedings of the 28th International Conference on Computational Linguistics, pages 198– 207, Barcelona, Spain (Online). International Committee o… find
  7. Zixiang Ding, Rui Xia, and Jianfei Yu. 2020. End-to-end emotion-cause pair extraction based on sliding window multi-label learning. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 3574–3583, Online. Association for Computational Linguistics. find
  8. Chuang Fan, Chaofa Yuan, Lin Gui, Yue Zhang, and Ruifeng Xu. 2021. Multi-task sequence tagging for emotion-cause pair extraction via tag distribution refinement. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 29:2339–2350. find
  9. Yahui Fu, Shogo Okada, Longbiao Wang, Lili Guo, Yaodong Song, Jiaxing Liu, and Jianwu Dang. 2021. Consk-gcn: Conversational semantic- and knowledge-oriented graph convolutional network for multimodal emotion recognition. In 2021 IEEE International Conference on Multimedia and Expo (ICME), pages 1–6. find
  10. Yao Fu, Shaoyang Yuan, Chi Zhang, and Juan Cao. 2023. Emotion recognition in conversations: A survey focusing on context, speaker dependencies, and fusion methods. Electronics, 12(22):4714. find
  11. Yingxue Gao, Huan Zhao, Yufeng Xiao, and Zixing Zhang. 2023. Gcformer: A graph convolutional transformer for speech emotion recognition. In IN-TERNATIONAL CONFERENCE ON MULTIMODAL INTERACTION, pages 307–313, Paris France. ACM. find
  12. Deepanway Ghosal, Navonil Majumder, Soujanya Poria, Niyati Chhaya, and Alexander Gelbukh. 2019. Dialoguegcn: A graph convolutional neural network for emotion recognition in conversation. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th Internation…
  13. Lin Gui, Dongyin Wu, Ruifeng Xu, Qin Lu, and Yu Zhou. 2016. Event-driven emotion cause extraction with corpus construction. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 1639–1649, Austin, Texas. Association for Computational Linguistics.
  14. Jingwen Hu, Yuchen Liu, Jinming Zhao, and Qin Jin. 2021. Mmgcn: Multimodal fusion via deep graph convolution network for emotion recognition in conversation. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natu… find
  15. DongJin Jeong and JinYeong Bak. 2023. Conversational emotion-cause pair extraction with guided mixture of experts. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, pages 3288–3298, Dubrovnik, Croatia. Association for Computational Lingui… find
  16. Xincheng Ju, Dong Zhang, Junhui Li, Shoushan Li, and Guodong Zhou. 2025. Enhanced generative framework with llms for multimodal emotion-cause pair extraction in conversations. IEEE Transactions on Multimedia, 27:4924–4935. find
  17. Sophia Yat Mei Lee, Ying Chen, and Chu-Ren Huang. 2010. A text-driven rule-based system for emotion cause detection. In Proceedings of the NAACL HLT 2010 Workshop on Computational Approaches to Analysis and Generation of Emotion in Text, pages 45–53, Los Angeles, CA. Association for Computational Li… find
  18. Bobo Li, Hao Fei, Fei Li, Tat-seng Chua, and Donghong Ji. 2024. Multimodal emotion-cause pair extraction with holistic interaction and label constraint. ACM Trans. Multimedia Comput. Commun. Appl. find
  19. Raymond Li, Samira Ebrahimi Kahou, Hannes Schulz, Vincent Michalski, Laurent Charlin, and Chris Pal. 2018. Towards deep conversational recommendations. In Advances in Neural Information Processing Systems, volume 31. Curran Associates, Inc. find
  20. Wei Li, Yang Li, Vlad Pandelea, Mengshi Ge, Luyao Zhu, and Erik Cambria. 2023. Ecpec: Emotion-cause pair extraction in conversations. IEEE Transactions on Affective Computing, 14(3):1754–1765. find
  21. Yanran Li, Hui Su, Xiaoyu Shen, Wenjie Li, Ziqiang Cao, and Shuzi Niu. 2017. Dailydialog: A manually labelled multi-turn dialogue dataset. In IJCNLP 2017.
  22. Tingting Liang, Chenxin Jin, Lingzhi Wang, Wenqi Fan, Congying Xia, Kai Chen, and Yuyu Yin. 2024. Llm-redial: A large-scale dataset for conversational recommender systems created from user behaviors with llms. In Findings of the Association for Computational Linguistics: ACL 2024, pages 8926–8939, B… find
  23. Bing Liu. 2022. Sentiment Analysis and Opinion Mining. Springer Nature. find
  24. Chen Liu, Changyong Niu, Jinge Xie, Yuxiang Jia, and Hongying Zan. 2023. Emotion-cause pair extraction in conversations based on multi-turn mrc with position-aware gcn. In 2023 International Conference on Asian Language Processing (IALP), pages 25–30. find
  25. Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. Roberta: A robustly optimized bert pretraining approach. Preprint, arXiv:1907.11692.
  26. Ilya Loshchilov and Frank Hutter. 2017. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101. find
  27. Junchi Ma, Hassan Nazeer Chaudhry, Farzana Kulsoom, Guihua Yang, Sajid Ullah Khan, Sujit Biswas, Zahid Ulalh Khan, and Faheem Khan. 2025. Multicausenet temporal attention for multimodal emotion cause pair extraction. Scientific Reports, 15. find
  28. Navonil Majumder, Soujanya Poria, Devamanyu Hazarika, Rada Mihalcea, Alexander Gelbukh, and Erik Cambria. 2019. Dialoguernn: An attentive rnn for emotion detection in conversations. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 6818–6825. find
  29. Hermina Petric Maretic and Mireille EL Gheche. 2019. Got: An optimal transport framework for graph comparison. In 33rd Conference on Neural Information Processing Systems (NeurIPS 2019). find
  30. Cam-Van Thi Nguyen, The-Son Le, Anh-Tuan Mai, and Duc-Trong Le. 2024. Ada2i: Enhancing modality balance for multimodal conversational emotion recognition. In Proceedings of the 32nd ACM International Conference on Multimedia, pages 9330– 9339, Melbourne VIC Australia. ACM. find
  31. Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, Alban Desmaison, Andreas Köpf, Edward Yang, Zach DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, Lu Fang, and 2 others. 2019.… — PyTorch: An Imperative Style, High-Performance Deep Learning Library
  32. Maria Pontiki, Dimitris Galanis, Haris Papageorgiou, Ion Androutsopoulos, Suresh Manandhar, Mohammad AL-Smadi, Mahmoud Al-Ayyoub, Yanyan Zhao, Bing Qin, Orphée De Clercq, Véronique Hoste, Marianna Apidianaki, Xavier Tannier, Natalia Loukachevitch, Evgeniy Kotelnikov, Nuria Bel, Salud María Jiménez-Z… — SemEval-2016 Task 5: Aspect Based Sentiment Analysis
  33. Soujanya Poria, Navonil Majumder, Devamanyu Hazarika, Deepanway Ghosal, Rishabh Bhardwaj, Samson Yu Bai Jian, Pengfei Hong, Romila Ghosh, Abhinaba Roy, Niyati Chhaya, Alexander Gelbukh, and Rada Mihalcea. 2021. Recognizing emotion cause in conversations. Cognitive Computation, 13(5):1317– 1332. find
  34. Hannah Rashkin, Eric Michael Smith, Margaret Li, and Y-Lan Boureau. 2019. Towards empathetic open-domain conversation models: A new benchmark and dataset. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 5370–5381, Florence, Italy. Association for Com… find
  35. Shruti Saxena and Joydeep Chandra. 2024. A survey on network alignment: Approaches, applications and future directions. In Proceedings of the Thirty-ThirdInternational Joint Conference on Artificial Intelligence, pages 8216–8224, Jeju, South Korea. International Joint Conferences on Artificial Intel… find
  36. Konstantinos Skitsas, Karol Orłowski, Judith Hermanns, Davide Mottin, and Panagiotis Karras. 2023. Comprehensive evaluation of unrestricted graph alignment algorithms. find
  37. Jianheng Tang, Kangfei Zhao, and Jia Li. 2023. A fused Gromov-Wasserstein framework for unsupervised knowledge graph entity alignment. In Findings of the Association for Computational Linguistics: ACL 2023, pages 3320–3334, Toronto, Canada. Association for Computational Linguistics. find
  38. Huynh Thanh Trung, Nguyen Thanh Toan, Tong Van Vinh, Hoang Thanh Dat, Duong Chi Thang, Nguyen Quoc Viet Hung, and Abdul Sattar. 2020. A comparative study on network alignment techniques. Expert Systems with Applications, 140:112883. find
  39. Petar Veličković, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Liò, and Yoshua Bengio. 2018. Graph attention networks. Preprint, arXiv:1710.10903.
  40. Botao Wang, Keke Tang, and Peican Zhu. 2024a. Enhancing emotion-cause pair extraction in conversations via center event detection and reasoning. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 10773–10783, Miami, Florida, USA. Association for Computational Linguistics… find
  41. Fanfan Wang, Zixiang Ding, Rui Xia, Zhaoyu Li, and Jianfei Yu. 2023a. Multimodal emotion-cause pair extraction in conversations. IEEE Transactions on Affective Computing, 14(3):1832–1844. find
  42. Yan Wang, Bo Wang, Yachao Zhao, Dongming Zhao, Xiaojia Jin, Jijun Zhang, Ruifang He, and Yuexian Hou. 2024b. Emotion recognition in conversation via dynamic personality. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-CO… find
  43. Yejiang Wang, Yuhai Zhao, Zhengkui Wang, and Ling Li. 2023b. Galopa: Graph transport learning with optimal plan alignment. In 37th Conference on Neural Information Processing Systems (NeurIPS 2023). find
  44. Rui Xia and Zixiang Ding. 2019. Emotion-cause pair extraction: A new task to emotion analysis in texts. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 1003– 1012, Florence, Italy. Association for Computational Linguistics. find
  45. Hongteng Xu, Dixin Luo, and Lawrence Carin. 2019. Scalable gromov-wasserstein learning for graph partitioning and matching. In Advances in Neural Information Processing Systems, volume 32. Curran Associates, Inc. find
  46. Zhichen Zeng, Boxin Du, Si Zhang, Yinglong Xia, Zhining Liu, and Hanghang Tong. 2024. Hierarchical multi-marginal optimal transport for network alignment. Proceedings of the AAAI Conference on Artificial Intelligence, 38(15):16660–16668. find
  47. Zhichen Zeng, Si Zhang, Yinglong Xia, and Hanghang Tong. 2023. Parrot: Position-aware regularized optimal transport for network alignment. In Proceedings of the ACM Web Conference 2023, pages 372–382, Austin TX USA. ACM. find
  48. Peixiang Zhong, Chen Zhang, Hao Wang, Yong Liu, and Chunyan Miao. 2020. Towards persona-based empathetic conversational models. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 6556–6566, Online. Association for Computational Linguistics.
  49. Peican Zhu, Botao Wang, Keke Tang, Haifeng Zhang, Xiaodong Cui, and Zhen Wang. 2024. A knowledge-guided graph attention network for emotion-cause pair extraction. Knowledge-Based Systems, 286:111342. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/ma-2026-emotion-cause-pair-extraction-conversations/