Hierarchical Shared Encoder With Task-Specific Transformer Layer Selection for Emotion-Cause Pair Extraction
Authors: 
Xinxin Su, Zhen Huang, Yixin Su, Bayu Distiawan Trisedya, Yong Dou, Yunxiang ZhaoAuthors Info & Claims
Xinxin Su, Zhen Huang, Yixin Su, Bayu Distiawan Trisedya, Yong Dou, Yunxiang ZhaoAuthors Info & ClaimsPages 1934 - 1948
Abstract
Emotion Cause Pair Extraction (ECPE) aims to extract emotions and their causes from a document. Powerful emotion and cause extraction abilities have proven essential in achieving accurate ECPE. However, most existing methods employ shared feature learning of emotion extraction and cause extraction, which can harm the abilities of both tasks as they focus on different information (i.e., task-specific features). Moreover, shared feature learning of the two tasks also leads to the label imbalance problem. To address these issues, this paper proposes a multi-task learning framework named Hierarchical Shared Encoder with Task-specific Transformer Layer Selection (HSE-TTLS). The model achieves ECPE via two subtasks: Emotion Extraction (EE) and Emotion Cause Extraction (ECE). The design of two subtasks for ECPE corresponds to the fact that cause clauses are emotion-dependent and significantly alleviates the label imbalance problem. To effectively extract task-specific features for EE and ECE, we employ BERT as the token-level encoder and select task-specific optimal layers for the two subtasks. Focal loss is used as the objective function for EE to further alleviate the label imbalance problem. Extensive experiments on benchmark ECPE corpus demonstrate the effectiveness of HSE-TTLS, which outperforms state-of-the-art baseline methods by at least 1.56% on the F1 score.
References
[1]
S. Y. M. Lee, Y. Chen, and C.-R. Huang, “A text-driven rule-based system for emotion cause detection,” in Proc. NAACL HLT Workshop Comput. Approaches Anal. Gener. Emotion Text, 2010, pp. 45–53.
[2]
L. Zhang, Y. Jia, X. Zhu, B. Zhou, and Y. Han, “User-level sentiment evolution analysis in microblog,” China Commun., vol. 11, no. 12, pp. 152–163, 2014.
[3]
J. Liu, M. Yan, and J. Luo, “Research on the construction of sentiment lexicon based on chinese microblog,” in Proc. 8th Int. Conf. Intell. Hum.- Mach. Syst. Cybern., 2016, pp. 56–59.
[4]
R. Xia and Z. Ding, “Emotion-cause pair extraction: A new task to emotion analysis in texts,” in Proc. 57th Annu. Meeting Assoc. Comput. Linguistics, 2019, pp. 1003–1012.
[5]
Z. Ding, R. Xia, and J. Yu, “ECPE-2D: Emotion-cause pair extraction based on joint two-dimensional representation, interaction and prediction,” in Proc. Annu. Meeting Assoc. Comput. Linguistics, 2020, pp. 3161–3170.
[6]
P. Wei, J. Zhao, and W. Mao, “Effective inter-clause modeling for end-to-end emotion-cause pair extraction,” in Proc. Annu. Meeting Assoc. Comput. Linguistics, 2020, pp. 3171–3181.
[7]
G. Hu, G. Lu, and Y. Zhao, “Bidirectional hierarchical attention networks based on document-level context for emotion cause extraction,” in Findings of the Association for Computational Linguistics. Stroudsburg, PA, USA: Association for Computational Linguistics, 2021, pp. 558–568.
[8]
X. Su, Z. Huang, Y. Zhao, Y. Chen, Y. Dou, and H. Pan, “Recent trends in deep learning based textual emotion cause extraction,” IEEE/ACM Trans. Audio Speech Lang. Process., vol. 31, pp. 2765–2786, 2023.
[9]
Z. Ding, R. Xia, and J. Yu, “End-to-end emotion-cause pair extraction based on sliding window multi-label learning,” in Proc. Conf. Empirical Methods Natural Lang. Process., 2020, pp. 3574–3583.
[10]
X. Chen, Q. Li, and J. Wang, “A unified sequence labeling model for emotion cause pair extraction,” in Proc. Int. Conf. Comput. Linguistics, 2020, pp. 208–218.
[11]
C. Zhou, D. Song, J. Xu, and Z. Wu, “A multi-turn machine reading comprehension framework with rethink mechanism for emotion-cause pair extraction,” in Proc. Int. Conf. Comput. Linguistics, 2022, pp. 6726–6735.
[12]
Z. Cheng, Z. Jiang, Y. Yin, H. Yu, and Q. Gu, “A symmetric local search network for emotion-cause pair extraction,” in Proc. Int. Conf. Comput. Linguistics, 2020, pp. 139–149.
[13]
Y. Chen, W. Hou, S. Li, C. Wu, and X. Zhang, “End-to-end emotion-cause pair extraction with graph convolutional network,” in Proc. Int. Conf. Comput. Linguistics, 2020, pp. 198–207.
[14]
Z. Cheng, Z. Jiang, Y. Yin, N. Li, and Q. Gu, “A unified target-oriented sequence-to-sequence model for emotion-cause pair extraction,” IEEE/ACM Trans. Audio Speech Lang. Process., vol. 29, pp. 2779–2791, 2021.
[15]
C. Yuan, C. Fan, J. Bao, and R. Xu, “Emotion-cause pair extraction as sequence labeling based on a novel tagging scheme,” in Proc. Conf. Empirical Methods Natural Lang. Process., 2020, pp. 3568–3573.
[16]
C. Fan, C. Yuan, L. Gui, Y. Zhang, and R. Xu, “Multi-task sequence tagging for emotion-cause pair extraction via tag distribution refinement,” IEEE/ACM Trans. Audio Speech Lang. Process., vol. 29, pp. 2339–2350, 2021.
[17]
Z. Cheng, Z. Jiang, Y. Yin, C. Wang, S. Ge, and Q. Gu, “A consistent dual-MRC framework for emotion-cause pair extraction,” ACM Trans. Inf. Syst., vol. 41, pp. 1046–8188, 2022.
[18]
J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “Bert: Pre-training of deep bidirectional transformers for language understanding,” in Proc. Conf. North Amer. Chapter Assoc. Comput. Linguistics, 2019, pp. 4171–4186.
[19]
Y. Chen, S. Y. M. Lee, S. Li, and C.-R. Huang, “Emotion cause detection with linguistic constructions,” in Proc. Int. Conf. Comput. Linguistics, 2010, pp. 179–187.
[20]
S. Y. M. Lee, Y. Chen, C.-R. Huang, and S. Li, “Detecting emotion causes with a linguistic rule-based approach 1,” Comput. Intell., vol. 29, no. 3, pp. 390–416, 2013.
[21]
L. Gui, L. Yuan, R. Xu, B. Liu, Q. Lu, and Y. Zhou, “Emotion cause detection with linguistic construction in chinese weibo text,” in Proc. CCF Int. Conf. Natural Lang. Process. Chin. Comput., 2014, pp. 457–464.
[22]
S. Song and Y. Meng, “Detecting concept-level emotion cause in microblogging,” in Proc. 24th Int. Conf. World Wide Web, 2015, pp. 119–120.
[23]
L. Gui, R. Xu, Q. Lu, D. Wu, and Y. Zhou, “Emotion cause extraction, a challenging task with corpus construction,” in Proc. Chin. Nat. Conf. Social Media Process., 2016, pp. 98–109.
[24]
X. Li, K. Song, S. Feng, D. Wang, and Y. Zhang, “A co-attention neural network model for emotion cause analysis with emotional context awareness,” in Proc. Conf. Empirical Methods Natural Lang. Process., 2018, pp. 4752–4757.
[25]
X. Yu, W. Rong, Z. Zhang, Y. Ouyang, and Z. Xiong, “Multiple level hierarchical network-based clause selection for emotion cause extraction,” IEEE Access, vol. 7, pp. 9071–9079, 2019.
[26]
R. Xia, M. Zhang, and Z. Ding, “Rthn: A RNN-transformer hierarchical network for emotion cause extraction,” in Proc. 28th Int. Joint Conf. Artif. Intell., 2019, pp. 5285–5291.
[27]
H. Yan, L. Gui, G. Pergola, and Y. He, “Position bias mitigation: A knowledge-aware graph model for emotion cause extraction,” in Proc. Annu. Meeting Assoc. Comput. Linguistics Int. Joint Conf. Natural Lang. Process., 2021, pp. 3364–3375.
[28]
C. Fan, C. Yuan, J. Du, L. Gui, and R. Xu, “Transition-based directed graph construction for emotion-cause pair extraction,” in Proc. Annu. Meeting Assoc. Comput. Linguistics, 2020, pp. 3707–3717.
[29]
H. Song and D. Song, “An end-to-end multi-task learning to link framework for emotion-cause pair extraction,” in Proc. Int. Conf. Image Video Process. Artif. Intell., 2021, pp. 13–21.
[30]
Y. Bao, Q. Ma, L. Wei, W. Zhou, and S. Hu, “Multi-granularity semantic aware graph model for reducing position bias in emotion cause pair extraction,” in Findings of the Association for Computational Linguistics. Stroudsburg, PA, USA: Association for Computational Linguistics, 2022, pp. 1203–1213.
[31]
W. Peng, Y. Hu, L. Xing, Y. Xie, Y. Sun, and Y. Li, “Control globally, understand locally: A global-to-local hierarchical graph network for emotional support conversation,” in Proc. Int. Joint Conf. Artif. Intell., 2022, pp. 4324–4330.
[32]
J. Gao et al., “Improving empathetic response generation by recognizing emotion cause in conversations,” in Findings of the association for computational linguistics. Stroudsburg, PA, USA: Association for Computational Linguistics, 2021, pp. 807–819.
[33]
A. S. Raamkumar and Y. Yang, “Empathetic conversational systems: A review of current advances, gaps, and opportunities,” IEEE Trans. Affective Comput., vol. 14, no. 4, pp. 2722–2739, Fourth Quarter 2022.
[34]
Y. Li et al., “Towards an online empathetic chatbot with emotion causes,” in Proc. 44th Int. ACM SIGIR Conf. Res. Develop. Inf. Retrieval, 2021, pp. 2041–2045.
[35]
S. Poria et al., “Recognizing emotion cause in conversations,” Cogn. Computation, vol. 13, pp. 1317–1332, 2021.
[36]
D. Zhang, Z. Yang, F. Meng, X. Chen, and J. Zhou, “TSAM: A two-stream attention model for causal emotion entailment,” in Proc. Int. Conf. Comput. Linguistics, 2022, pp. 6762–6772.
[37]
A. Bhat and A. Modi, “Multi-task learning framework for extracting emotion cause span and entailment in conversations,” in Proc. Transfer Learn. Natural Lang. Process. Workshop, 2023, pp. 33–51.
[38]
A. B. Zadeh, P. P. Liang, S. Poria, E. Cambria, and L.-P. Morency, “Multimodal language analysis in the wild: CMU-mosei dataset and interpretable dynamic fusion graph,” in Proc. 56th Annu. Meeting Assoc. Comput. Linguistics, 2018, pp. 2236–2246.
[39]
C. Busso et al., “Iemocap: Interactive emotional dyadic motion capture database,” Lang. Resour. Eval., vol. 42, pp. 335–359, 2008.
[40]
A. Joshi, A. Bhat, A. Jain, A. Singh, and A. Modi, “COGMEN: Contextualized GNN based multimodal emotion recognition,” in Proc. Conf. North Amer. Chapter Assoc. Comput. Linguistics: Hum. Lang. Technol., 2022, pp. 4148–4164.
[41]
K. Bansal, H. Agarwal, A. Joshi, and A. Modi, “Shapes of emotions: Multimodal emotion recognition in conversations via emotion shifts,” in Proc. 1st Workshop Perform. Interpretability Evaluations Multimodal Multipurpose, Massive-Scale Models, 2022, pp. 44–56.
[42]
T. Shi and S.-L. Huang, “MultiEMO: An attention-based correlation-aware multimodal fusion framework for emotion recognition in conversations,” in Proc. Annu. Meeting Assoc. Comput. Linguistics, 2023, pp. 14752–14766.
[43]
D. Zhang, F. Chen, and X. Chen, “Dualgats: Dual graph attention networks for emotion recognition in conversations,” in Proc. Annu. Meeting Assoc. Comput. Linguistics, 2023, pp. 7395–7408.
[44]
S. Ghosh, M. Chollet, E. Laksana, L.-P. Morency, and S. Scherer, “Affect-LM: A neural language model for customizable affective text generation,” in Proc. Annu. Meeting Assoc. Comput. Linguistics, 2017, pp. 634–642.
[45]
P. Colombo, W. Witon, A. Modi, J. Kennedy, and M. Kapadia, “Affect-driven dialog generation,” in Proc. Conf. North Amer. Chapter Assoc. Comput. Linguistics: Hum. Lang. Technol., 2019, pp. 3734–3743.
[46]
T. Goswamy, I. Singh, A. Barkati, and A. Modi, “Adapting a language model for controlled affective text generation,” in Proc. 28th Int. Conf. Comput. Linguistics, 2020, pp. 2787–2801.
[47]
Y. M. Resendiz and R. Klinger, “Emotion-conditioned text generation through automatic prompt optimization,” 2023,.
[48]
G. Jawahar, B. Sagot, and D. Seddah, “What does BERT learn about the structure of language?,” in Proc. Annu. Meeting Assoc. Comput. Linguistics, 2019, pp. 3651–3657.
[49]
R. T. Mccoy, T. Linzen, E. Dunbar, and P. Smolensky, “RNNs implicitly implement tensor product representations,” in Proc. Int. Conf. Learn. Representations, 2019.
[50]
S.-M. Wang and L.-W. Ku, “Antusd: A large chinese sentiment dictionary,” in Proc. 10th Int. Conf. Lang. Resour. Eval., 2016, pp. 2697–2702.
[51]
Y. Li, X. Tian, T. Liu, and D. Tao, “Multi-task model and feature joint learning,” in Proc. 24th Int. Joint Conf. Artif. Intell., 2015, pp. 3643–3649.
[52]
A. Singh, S. Hingane, S. Wani, and A. Modi, “An end-to-end network for emotion-cause pair extraction,” in Proc. Eleventh Workshop Comput. Approaches Subjectivity Sentiment Social Media Anal., 2021, pp. 84–91.
[53]
Z. Wang, Q. Xie, Z. Ding, Y. Feng, and R. Xia, “Is chatgpt a good sentiment analyzer? A preliminary study,” 2023,.
[54]
J. Wu, Y. Shen, Z. Zhang, and L. Cai, “Enhancing large language model with decomposed reasoning for emotion cause pair extraction,” 2024,.
[55]
C. Li, J. Hu, T.-J. Li, S. Du, and F. Teng, “An effective multi-task learning model for end-to-end emotion-cause pair extraction,” Appl. Intell., vol. 53, pp. 3519–3529, 2022.
[56]
S. Chen et al., “Joint alignment of multi-task feature and label spaces for emotion cause pair extraction,” in Proc. Int. Conf. Comput. Linguistics, 2022, pp. 6955–6965.
[57]
J. Liu et al., “Emotion-cause pair extraction with bidirectional multi-label sequence tagging,” Appl. Intell., vol. 53, pp. 30400–30415, 2023.
[58]
W. Zhao, Y. Zhao, Z. Li, and B. Qin, “Knowledge-bridged causal interaction network for causal emotion entailment,” 2022,.
[59]
J. Li et al., “Neutral utterances are also causes: Enhancing conversational causal emotion entailment with social commonsense knowledge,” in Proc. Int. Joint Conf. Artif. Intell., 2022, pp. 4209–4215.
[60]
S. Yoo and O. Jeong, “A token classification-based attention model for extracting multiple emotion–cause pairs in conversations,” Sensors, vol. 23, 2023, Art. no.
[61]
I. Loshchilov and F. Hutter, “Fixing weight decay regularization in adam,” 2017, arXiv:1711.05101.
[62]
Z. Liu, W. Lin, Y. Shi, and J. Zhao, “A robustly optimized bert pre-training approach with post-training,” in Proc. 20th Chin. Nat. Conf. Comput. Linguistics, 2021, pp. 471–484.
[63]
A. C. Stickland and I. Murray, “Bert and pals: Projected attention layers for efficient adaptation in multi-task learning,” in Proc. Int. Conf. Mach. Learn., 2019, pp. 5986–5995.
[64]
C. Fan et al., “A knowledge regularized hierarchical approach for emotion cause analysis,” in Proc. Conf. Empirical Methods Natural Lang. Process. Int. Joint Conf. Natural Lang. Process., 2019, pp. 5614–5624.
[65]
X. Li, S. Feng, Y. Zhang, and D. Wang, “Multi-level emotion cause analysis by multi-head attention based multi-task learning,” China Nat. Conf. Chin. Comput. Linguistics, 2021, pp. 77–93.
Index Terms
- Hierarchical Shared Encoder With Task-Specific Transformer Layer Selection for Emotion-Cause Pair Extraction
Index terms have been assigned to the content through auto-classification.
Recommendations
An effective multi-task learning model for end-to-end emotion-cause pair extraction
AbstractEmotion-cause pair extraction (ECPE), as an extended research direction of emotion cause extraction, aims to extract emotion and its corresponding causes for a given document. Previous methods solved this problem in a two-stage fashion. ...
Emotion Cause Pair Extraction Based on Multitask
Web Information Systems and ApplicationsAbstractEmotion cause pair extraction is a sub-task of sentiment analysis tasks, which aims to extract all emotion clauses in a given document and the cause clauses corresponding to the emotion. At present, only the method of sharing parameters at the ...
Unifying emotion-oriented and cause-oriented predictions for emotion-cause pair extraction
AbstractEmotion-cause pair extraction (ECPE) is an extraction task aiming to simultaneously identify the emotions and causes from the text without emotion annotations. Let c i and c j represent the emotion clause and the cause clause of a document, ...
Comments
Information & Contributors
Information
Published In
© 2024 The Authors. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/.
Publisher
IEEE Computer Society Press
Washington, DC, United States
Publication History
Published: 17 April 2024
Qualifiers
- Research-article
Contributors
Other Metrics
Bibliometrics & Citations
Bibliometrics
Article Metrics
- 0Total Citations
- 0Total Downloads
- Downloads (Last 12 months)0
- Downloads (Last 6 weeks)0
Reflects downloads up to 25 Dec 2024