Reproduce “PsyChat: A Client-Centric Dialogue System for Mental Health Support” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Huachuan Qiu, Anqi Li, Lizhi Ma, Zhenzhong Lan
Researchers from Zhejiang University and Westlake University developed PsyChat, a client-centric dialogue system for mental health support that dynamically tailors counseling strategies to recognized client behaviors. This system generated responses judged by professional counselors as statistically comparable in quality to human-provided ground truth, while also demonstrating improved fluency, relevance, and diversity over a baseline model.
Key insights
- Explicitly integrating client behavior recognition into the dialogue system's architecture enables more adaptive and personalized mental health support, mimicking the dynamic adjustments made by human counselors.
- Combining dense retrieval with a demonstration-based approach for counselor strategy selection ensures that the system's responses are contextually relevant and aligned with identified client behaviors.
- A two-stage fine-tuning process, utilizing both large-scale pseudo-labeled data and authentic real-world counseling data, effectively adapts large language models to generate high-quality, specialized responses for mental health dialogue.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.