Annual Conference of KIPS (한국정보처리학회:학술대회논문집)

  • 2024.10a
  • /
  • Pages.511-514
  • /
  • 2024
  • /
  • 2005-0011(pISSN)
  • /
  • 2671-7298(eISSN)
Korea Information Processing Society (한국정보처리학회)
권호 표지
DOI QR코드

DOI QR Code

Optimizing Empathy Prediction in Text-Based Mental Health Support with Multi-Task Learning and Imbalance Mitigation

  • Anjitha Divakaran (Dept. of AI Convergence, Chonnam National University) ;
  • Hyung-Jeong Yang (Dept. of AI Convergence, Chonnam National University) ;
  • Seung-won Kim (Dept. of AI Convergence, Chonnam National University) ;
  • Ji-eun Shin (Dept. of Psychology, Chonnam National University) ;
  • Soo-Hyung Kim (Dept. of AI Convergence, Chonnam National University)
  • Published : 2024.10.31
Download PDF
⟨ Previous Next ⟩

Abstract

Empathy plays a crucial role in effective mental health support, particularly on text-based platforms where individuals seek understanding and compassion. Building on previous work that predicted empathy in separate models, we propose a multi-task learning approach that jointly models three empathy communication mechanisms: emotional reactions, interpretations, and explorations. By integrating RoBERTa with GRU layers, our model improves the accuracy and efficiency of empathy detection. Additionally, we address class imbalance by incorporating focal loss, which helps the model focus on underrepresented strong empathy signals. Our experiments show that the multi-task model, combined with focal loss, significantly improves performance across all empathy dimensions, making it a promising tool for enhancing real-time empathy feedback in online mental health support systems.

Keywords

Acknowledgement

This research was supported by the Ministry of Science and ICT (MSIT), Korea, under the Innovative Human Resource Development for Local Intellectualization support program (IITP-2023-RS-2022-00156287) supervised by the Institute for Information & communications Technology Planning & Evaluation (IITP). Additionally, it was funded by IITP under the Artificial Intelligence Convergence Innovation Human Resources Development (IITP-2023-RS-2023-00256629) grant, and by the National Research Foundation of Korea (NRF) through the Korea government (MSIT) grant (RS2023-00219107).

References

  1. Dhyani A, Gaidhane A, Choudhari SG, Dave S, Choudhary S. Strengthening Response Toward Promoting Mental Health in India: A Narrative Review. Cureus. 2022 Oct 18;14(10):e30435. doi: 10.7759/cureus.30435. PMID: 36407164; PMCID: PMC9671264.
  2. Webb, M., Burns, J., & Collin, P. (2008). Providing online support for young people with mental health difficulties: challenges and opportunities explored. Early intervention in psychiatry, 2(2), 108-113.
  3. Ashish Sharma, Adam Miner, David Atkins, and Tim Althoff. 2020. A Computational Approach to Understanding Empathy Expressed in Text-Based Mental Health Support. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 5263-5276, Online. Association for Computational Linguistics.
  4. Veronica Perez-Rosas, Rada Mihalcea, Kenneth Resnicow, Satinder Singh, and Lawrence An. 2017. Understanding and Predicting Empathic Behavior in Counseling Therapy. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1426-1435, Vancouver, Canada. Association for Computational Linguistics.
  5. Buechel, S., Buffone, A., Slaff, B., Ungar, L.H., & Sedoc, J. (2018). Modeling Empathy and Distress in Reaction to News Stories. Conference on Empirical Methods in Natural Language Processing.
  6. 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 Computational Linguistics.
  7. Ronan Collobert, Jason Weston, Leon Bottou, Michael Karlen, Koray Kavukcuoglu, and Pavel Kuksa. 2011. Natural Language Processing (Almost) from Scratch. J. Mach. Learn. Res. 12, null (2/1/2011), 2493-2537.
  8. Xiaodong Liu, Pengcheng He, Weizhu Chen, and Jianfeng Gao. 2019. Multi-Task Deep Neural Networks for Natural Language Understanding. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 4487-4496, Florence, Italy. Association for Computational Linguistics.
  9. Ruder, S. (2017). An Overview of Multi-Task Learning in Deep Neural Networks. ArXiv, abs/1706.05098.
  10. H. He and E. A. Garcia, "Learning from Imbalanced Data," in IEEE Transactions on Knowledge and Data Engineering, vol. 21, no. 9, pp. 1263-1284, Sept. 2009, doi: 10.1109/TKDE.2008.239.
  11. T. -Y. Lin, P. Goyal, R. Girshick, K. He and P. Dollar, "Focal Loss for Dense Object Detection," 2017 IEEE International Conference on Computer Vision (ICCV), Venice, Italy, 2017, pp. 2999-3007, doi: 10.1109/ICCV.2017.324.
  12. Usuga-Cadavid, J., Gil, D., & Ruiz, E. (2021). Exploring the influence of focal loss on transformer models. Procedia Computer Science, 184, 193-200. https://doi.org/10.1016/j.procs.2021.03.055