GitHub - SALT-NLP/CARE: All code of CARE: model training, frontend, backend, and analysis (original) (raw)

CARE: Helping the Helper

This repository contains the codebase for the paper:

Helping the helper: Supporting peer counselors via ai-empowered practice and feedbackHsu, Shang-Ling, Raj Sanjay Shah, Prathik Senthil, Zahra Ashktorab, Casey Dugan, Werner Geyer, and Diyi Yang.Proceedings of the ACM on Human-Computer Interaction 9, no. 2 (2025): 1-45.Read the paper here

CARE is a system designed to support peer counselors by providing AI-empowered practice and feedback. It helps counselors diagnose which skills to use and suggests possible responses during interactions with support seekers.

Repository Structure

• analysis/: Notebooks and scripts for analyzing conversation logs and user profiles.
• backend/: Flask-based API that serves the models and manages the chat sessions.
• frontend/: React-based web interface for the counseling chat.
• models/: Scripts and notebooks for training the response generation and skill prediction models.

Getting Started

Prerequisites

• Python 3.9+
• Node.js and npm
• Conda (recommended for environment management)

Installation

1. Clone the repository:
   git clone
   cd care
2. Setup the Backend:
   cd backend
   pip install -r requirements.txt
3. Setup the Frontend:

Usage

1. Model Training

Before running the system, you must train your own models. Navigate to the models/ directory for scripts to train the skill predictors and response generators.

• Use predict_next_code.py to train the MITI code predictor.
• Use generate_next_utterance.py to train the response generation model.

Note: You will need to update the data paths in these scripts to point to your training datasets.

2. Running the Backend

Once models are trained, update the paths in backend/api.py to point to your model checkpoints.

Run the backend server:

3. Running the Frontend

Start the React development server:

The interface should now be accessible at http://localhost:3000.

Citation

If you use this codebase or CARE in your research, please cite our paper:

@article{hsu2025helping, title={Helping the helper: Supporting peer counselors via ai-empowered practice and feedback}, author={Hsu, Shang-Ling and Shah, Raj Sanjay and Senthil, Prathik and Ashktorab, Zahra and Dugan, Casey and Geyer, Werner and Yang, Diyi}, journal={Proceedings of the ACM on Human-Computer Interaction}, volume={9}, number={2}, pages={1--45}, year={2025}, publisher={ACM New York, NY, USA} }

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contact

For any questions, please contact Shang-Ling Hsu at hsushang@usc.edu.