PsyQA: A Chinese Dataset for Generating Long Counseling Text for Mental Health Support

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 1489–1503 August 1–6, 2021. ©2021 Association for Computational Linguistics 1489 PsyQA: A Chinese Dataset for Generating Long Counseling Text for Mental Health Support Hao Sun1∗, Zhenru Lin2∗, Chujie Zheng1, Siyang Liu3, Minlie Huang1† 1The CoAI group, DCST, Institute for Artificial Intelligence, State Key Lab of Intelligent Technology and Systems, 1Beijing National Research Center for Information Science and Technology, Tsinghua University, Beijing 100084, China 2Dept. of Electronic Engineering, Tsinghua University, Beijing 100084, China 3Tsinghua-Berkeley Shenzhen Institute, Tsinghua Shenzhen International Graduate School, 3Tsinghua University, Shenzhen, China {h-sun20,linzr18}@mails.tsinghua.edu.cn, aihuang@tsinghua.edu.cn Abstract Great research interests have been attracted to devise AI services that are able to provide men- tal health support. However, the lack of cor- pora is a main obstacle to this research, par- ticularly in Chinese language. In this paper, we propose PsyQA, a Chinese dataset of psy- chological health support in the form of ques- tion and answer pair. PsyQA is crawled from a Chinese mental health service platform, and contains 22K questions and 56K long and well- structured answers. Based on the psychologi- cal counseling theories, we annotate a portion of answer texts with typical strategies for pro- viding support, and further present in-depth analysis of both lexical features and strategy patterns in the counseling answers. We also evaluate the performance of generating coun- seling answers with the generative pretrained models. Results show that utilizing strategies enhances the fluency and helpfulness of gener- ated answers, but there is still a large space for future research. 1 Introduction The burden of mental disorders continues to grow with significant impacts on human health and social development (Organization et al., 2011; The World Health Organization, 2020). As an effective therapy for mental disorders (Reynolds Jr et al., 2013), on- line mental health counseling, which mostly refers to communicating anonymously, has become popu- lar in recent years (Fu et al., 2020). Great research interests have been endeavored to devise AI services that are able to provide men- tal health support (Bucci et al., 2019; Liu et al., 2021). Based on the online-text psychotherapy corpora, previous works have utilized text min- ing techniques to detect empathy (Sharma et al., 2020; Zheng et al., 2021), linguistic development of ∗Equal contribution. †Corresponding author. counselors (Zhang et al., 2019), and self-injurious thoughts and behaviors (Franz et al., 2020). How- ever, the research of text-based mental health coun- seling is still largely limited due to the lack of relevant corpora, particularly in Chinese language. To this end, we collect PsyQA in this work, a Chinese dataset of Psychological health support in the form of Question-Answer pair. An exam- ple data of PsyQA is shown in Figure 1. In each example, the question along with a detailed de- scription and several keyword tags is posted by an anonymous help-seeker, where the description generally contains dense persona and emotion in- formation about the help-seeker. The answer is usually quite long (524 words on average). The answers are replied asynchronously from the well- …

Content truncated to 500 words