bBot Rasa.ai NLU adapter

NLU adapter for analysing bBot messages with Rasa.ai.

~(O_O)~

Visit bBot.chat for info on the conversation engine.

Connecting to Rasa

Local NLU Starter

If you don't already have a Rasa instance, you can use the Rasa starter pack, which runs a very basic pre-trained NLU API locally, for testing interactions.

Alternatively, within ./sample is the Rasa quickstart content, if you would rather build your own from scratch.

Use the package script to clone the Rasa NLU starter pack:

yarn rasa:starter

If you haven't installed Rasa NLU yet, run:

yarn rasa:install

Train the NLU instance with the starter data:

yarn rasa:train

Run the NLU server:

yarn rasa:run

You can now make HTTP requests to your local server:

curl XPOST localhost:5000/parse -d '{"query":"Hello", "project": "current"}'

Configuring the bot

The default configs (as per below) can be modified for your custom Rasa hosts.

Add these lines to the bot environment or .env file in local path for testing.

BOT_RASA_URL=localhost:5000
BOT_RASA_PROJECT=current

Also set BOT_NLU_MIN_LENGTH=2 if you want NLU to process like 'hi' and 'bye'

These options can also be given as CLI args like --rasa-url localhost:5000 or as attributes in package.json under "bot" like "rasa-project": "current".

Contributing

This adapter is functional but only provides MVP utility and needs to be matured with more "real world" experience. If you're using Rasa in production with more advanced requirements, like custom extractors and actions, please report an issue to describe your usage and requirements, so we might improve it together.