Beta - Please keep in mind that this application is under heavy development

Rasa UI

Rasa UI is a web application built on top of, and for, Rasa NLU. Rasa UI provides a web application to quickly and easily be able to create agents, define intents and entities. It also provides some convenience features for Rasa NLU, like training your models, monitoring usage or viewing logs. Our goal is to replace API.ai with Rasa, so a lot of the terminology and usage concepts are similar.

Features

• Training data stored in DB
• UI for managing training data
• Initiate training from UI
• Review Rasa configuration and component pipelines
• Log requests for usage tracking, history, improvements
• Usage dashboard
• Easily execute intent parsing using different models

Getting Started

Rasa UI can run directly on your Rasa NLU instance, or on a separate machine. Technically Rasa NLU is not required, you could just use the UI for managing training data.

Prerequisites

Rasa NLU - Version 8.2.?+

PostgreSQL - Used for storing training data (entities, intents, synonyms, etc.)

Node.js/npm - Serves Rasa UI and acts as a middleware server for logging (to the PostgreSQL DB)

Installing

Please ensure prerequisites are fulfilled

Clone/download the Rasa UI repository

git clone https://github.com/paschmann/rasaui.git

npm install

Please see the wiki for more detailed instructions.

Running

Run npm start from the server folder

npm start

• Update your package.json file to include the IP Addresses of your rasa server and the connection string of your postgres instance.
• Update your web/src/app.js file to include the IP Addresses of your local middleware server (no need to change this if they are running on the same instance)

Your web application should be available on http://localhost:5001

Logging

Since Rasa UI can be used to log events/intent parsing/training etc. we would suggest changing your endpoints for your API calls to "pass through" the Rasa UI middleware layer. All API requests are simply forwarded, logged and then returned.

e.g. Instead of calling: http://localhost:5000/parse?q=hello%20there rather call: http://localhost:5001/api/parse?q=hello%20there

Contributing

Please read contributing.md for details on our code of conduct, and the process for submitting pull requests to us.

Authors

• Paul Aschmann

See also the list of contributors who participated in this project.

License

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