mental_health_chatbot

Implements a lightweight NLU Server using RASA & FastAPI with a Streamlit Interface

Project Requirements

1. Install Poetry i.e. curl -sSL https://install.python-poetry.org | python3 -
2. Make sure you have a compatible version of python as specified in the pyproject.toml
3. Run poetry install in your environment or create a poetry environment

Running the Application

Run the following from the terminal (make sure the relevant ports aren't already in use)

./start.sh

This should open the streamlit app in a new browser window where you can interact with the Intent Detection Model

Calling the FastAPI Endpoint from Another Service

Once the start script has been run the FastAPI app should be up and running at http://0.0.0.0:8000 and can be called via a POST request:

curl -X POST -H "Content-Type: application/json" -d '{"text":"hello"}' http://localhost:8000/parse/

The response will contain the name of the detected intent, the confidence associated with the prediction and the integer value mapped to the intent:

{"intent":"greet","confidence":0.9999904632568359,"intent_number":6}

Model Evaluation

Run rasa test nlu --nlu train_test_split/test_data.yml and the results should be visible in the results folder

Changing the Model

If you have trained an alternative model you can change the model name in the start.sh script and restart.