Abstract
Emotional cause identification (ECI) is an important task for emotion analysis, aiming to identify the causes behind a certain emotion expressed in the text. Most of the previous studies are restricted to the clause-level binary classification, which have ignored an important fact that not all words in the clause are useful cause information for people to express emotions. In this work, we propose a new task: emotion-cause span extraction (ECSE), which is capable of obtaining more accurate and effective emotion causes. Inspired by recent advances in using joint learning approaches to ECI, we propose a novel joint learning framework for emotion-cause span extraction and span-based emotion classification so as to better address the ECSE task. Taken as the default backbone network, the Bidirectional Encoder Representations from Transformers (BERT) is used to encode multiple words and serve contextualized token representations. Furthermore, we also propose a multi-attention mechanism with emotional context awareness and a relative position learning mechanism on word-level, which is able to further capture the mutual interactions between the emotion clauses and candidate spans. According to the experimental results on a benchmark emotion cause corpus, it proves the reliability of the ECSE task and the effectiveness of our approach. In addition, through an in-depth analysis of traditional ECI task by converting ECSE into the clause-level binary classification task, we achieve the best performance among the systems in comparison, which further demonstrates the feasibility of the new ECSE task.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.Notes
The corpus was published by Gui [1] in 2017
ELMO model is adapted from this implementation: https://github.com/HIT-SCIR/ELMoForManyLangshttps://github.com/HIT-SCIR/ELMoForManyLangs. In ECSE-ELMO method, we set the embedding dimensionality to be d = 768. The initial learning rate is 105 and the batch size is 32
GPT model is adapted from this implementation: https://github.com/openai/gpt-2. In ECSE-GPT method, we use the learned embeddings with supported sequence lengths of up to 512 tokens to be consistent with our experimental setting. A batch size of 8, learning rate is 5e-5 and classifier dropout with a rate of 0.1
References
Gui L, Wu D, Xu R et al (2016) Event-driven emotion cause extraction with corpus construction. In: Proceedings of the 2016 conference on empirical methods in natural language processing, pp 1639–1649
Cheng X, Chen Y, Cheng B et al (2017) An emotion cause corpus for chinese microblogs with multiple-user structures, vol 17
Yada S, Ikeda K, Hoashi K et al (2017) A bootstrap method for automatic rule acquisition on emotion cause extraction. In: 2017 IEEE international conference on data mining workshops (ICDMW), IEEE, pp 414–421
Ding Z, He H, Zhang M et al (2019) From independent prediction to reordered prediction: Integrating relative position and global label information to emotion cause identification. In: Proceedings of the AAAI conference on artificial intelligence (AAAI)
Chen Y, Hou W, Cheng X et al (2018) Joint learning for emotion classification and emotion cause detection. In: Proceedings of the 2018 conference on empirical methods in natural language processing (EMNLP), pp 646–651
Xu R, Hu J, Lu Q, at al (2017) An ensemble approach for emotion cause detection with event extraction and multi-kernel svms. Tsinghua Sci Technol 22(6):646–659
Lee S, Y M, Chen Y, Huang CR, at al (2013) Detecting emotion causes with a linguistic rulebased approach, vol 29
Li W, Xu H (2014) Text-based emotion classification using emotion cause extraction. Expert Syst Appl 41(4):1742–1749
Ghazi D, Inkpen D, Szpakowicz S (2015) Detecting emotion stimuli in emotion-bearing sentences. In: International conference on intelligent text processing and computational linguistics, Springer, pp 152–165
Gui L, Hu J, He Y, Xu R, Lu Q, Du J (2017) A question answering approach to emotion cause extraction. In: Empirical methods in natural language processing (EMNLP), pp 1593–1602
Chen Y, Hou W, Cheng X et al (2018) Hierarchical convolution neural network for emotion cause detection on microblogs. In: International conference on artificial neural networks (ICANN), pp 115–122
Xu B, Lin H, Li Y et al (2019) Extracting emotion causes using learning to rank methods from an information retrieval perspective, vol 7
Xia R, Ding Z (2019) Emotion-cause pair extraction: A new task to emotion analysis in texts. In: Proceedings of the 57th conference of the association for computational linguistics, ACL 2019, Florence, Italy, July 28- August 2, 2019, Vol 1: Long Papers, pp 1003–1012
Fan C, Yuan C, Du J et al (2020) Transition-based directed graph construction for emotion-cause pair extraction[C]// Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics
Gao W, Li S, Lee S, at al (2013) Joint learning on sentiment and emotion classification. In: Proceedings of the 22nd ACM international conference on information & knowledge management, pp 1505–1508
Wang R, Li S, Zhou D, at al (2015) Joint sentiment and emotion classification with integer linear programming. In: Liu A, Ishikawa Y, Qian T, Nutanong S, Cheema M (eds) Database systems for advanced applications, lecture notes in computer science, Vol 9052
You Q (2016) Sentiment and emotion analysis for social multimedia: methodologies and Applications[C]// the 2016 ACM ACM
Pavitra R, Kalaivaani PCD (2015) Weakly supervised sentiment analysis using joint sentiment topic detection with bigrams[C]// 2015 2nd International Conference on Electronics and Communication Systems (ICECS) IEEE
Huang F, Zhang S, Zhang J et al Multimodal learning for topic sentiment analysis in microblogging[J]. Neurocomputing, 2017:S0925231217304393
Li W, Xu H (2014) Text-based emotion classification using emotion cause extraction. Expert Syst Appl 41(4):1742–1749
Gao K, Xu H, Wang J (2015) Emotion cause detection for chinese micro-blogs based on ecocc model. In: Pacific-Asia conference on knowledge discovery and data mining, Springer, pp 3–14
Li J, Song S, Feng S, Wang L, at al (2018) A co-attention neural network model for emotion cause analysis with emotional context awareness. In: Proceedings of the 2018 conference on empirical methods in natural language processing (EMNLP), pp 4752–4757
Devlin J, Chang W, Lee T, at al (2018) Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv:1810.04805
Xia R, Zhang M, Ding Z (2019) Rthn: a rnn-transformer hierarchical network for emotion cause extraction
Vaswani A, Shazeer N, Parmar N (2017) Attention is all you need. Inproceedings of NIPS
Zhang H, Qi P, D C (2018) Graph convolution over pruned dependency trees improves relation extraction. In: Proceedings of the 2018 conference on empirical methods in natural language processing (EMNLP)
Hu M, Peng Y, Huang Z et al (2019) Open-domain targeted sentiment analysis via span-based extraction and classification
Bahdanau D, Cho K, Bengio Y (2014) Neural machine translation by jointly learning to align and translate. arXiv:1409.0473
Peters M, Neumann M, Iyyer M (2018) Deep Contextualized Word Representations[J]
Alec R, Jeffrey W, Rewon C, David L et al (2019) Language models are unsupervised multitask learners. Technical report, OpenAI
Lee S Y M, Chen Y, Huang CR (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. Association for Computational Linguistics
Russo I, Caselli T, Rubino F, et al. (2011) EMOCause: an easy-adaptable approach to emotion cause contexts[C]// Workshop on Computational Approaches to Subjectivity & Sentiment Analysis. Association for Computational Linguistics
Acknowledgements
This work was supported by the National Natural Science Foundation of China (grant number 61561047); Xin Jiang education funds project of china (grant number 90390007); Key Laboratory project of Xinjiang Normal University, China (grant number XJNUSYS102018B04).
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Li, M., Zhao, H., Su, H. et al. Emotion-cause span extraction: a new task to emotion cause identification in texts. Appl Intell 51, 7109–7121 (2021). https://doi.org/10.1007/s10489-021-02188-7
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10489-021-02188-7