CovidBot-RASA

RASA based covid bot on flask

Python environment

• Create a python virtual environment

• Activate the virtual environment

• All the commands to be run in the virtual environment

  python3 -m venv rasa_env
  rasa_env\Scripts\activate.bat #On Windows
  source rasa-env/bin/activate #On nix based

Rasa set up

Install rasa

• Ensure you are on the Python virtual environment

• Install rasa pip install rasa

• Alternatively, execute the requirements.txt from the repo

  pip install -r requirements.txt

Initiate rasa

• Clone this project if you don't want to initiate rasa on your own.
• Alternatively, create an empty directory where you want to the project to live
• Cd into your project directory
• Command to initiate rasa : rasa init
• This will create all the rasa related files with basic content

Add Covid specific intents and entities to data/nlu.md

• corona_state is a project specific intents
• Other intents comes with rasa init

Add actions in data/stories.md

• Add a new story to connect the new intent corona_state to new action
• new action name is action_corona_tracker

Add action in domain.yml

• Add a new action : action_corona_tracker

Ensure endpoints in endpoints.yml

• Ensure the endpoints are uncommented in endpoints.yml
• action_endpoint points to localhost at port 5055

Add custom actions to actions.py

• Actual logic of the tracker happens here.
• Tracker has the inputs to the custom function
• Dispatcher gives back the output

Test and interact with your bot

• On a new terminal execute command rasa shell
• Interact and test your bot on command line.

Flask set up

Flask entry file : app.py

• If you had executed requirements.txt, flask should be already there
• Else install it using pip install flask
• app.py is the flask entry point
• app.py interacts with the rasa endpoint
• app.py renders the output to a html page

UI page at templates/index.html

• Accepts simple text input from app.py

How to run the set up

Train rasa

• Run command rasa train at the project folder
• This should create a new model file under models directory
• Train and regenerate whenever changes to rasa specific files are made

Create an action server

• Custom actions require a seperate server to execute on
• Command to create action server rasa run actions
• Keep the action terminal open to allow the action server to live

Run rasa

• Open a new terminal and run rasa with api
• rasa run -m models --enable-api
• Keep this rasa terminal open as well

Run flask

• To run flask execute run flask