Reputation: 524
I am planning to build a chat-bot using RASA with Node.js as backend. The challenge is, the chat-bot is to be housed on intranet network(i.e. it should work completely offline). What should be my approach to achieve this? I have tried integration of Bot-kit framework with RASA, but this brings about a dependency on internet(need to register your app on the website).
Upvotes: 2
Views: 2154
Reputation: 5853
RASA NLU will only help in you intent classification and entity recognition. For fulfilling after that upon your business logic you need an appropriate framework to do that.
Currently, as of now there's no robust framework for Chatbot development which completely is independent from the internet and on-premise.
However you can use the Microsoft Bot Framework SDK for Node.JS for fulfilling your response coupled with RASA NLU for Intent classification and Entity detection. The Bot framework gives you the functionality to add support to your chatbots to multiple channels like Skype, Facebook messenger, Slack, Web Chat, etc.
I forgot to mention, you can also use RASA Core for fulfilling your business logic based on slots and stories. Its a machine learning model and doesn't require complex state machines like Microsoft Bot Framework and its completely on-premise.
Upvotes: 0
Reputation: 1832
We have an opensource project to address that specific need.
The project is called Articulate.
Its currently in the alpha/beta phase, but its a (node based) wrapper around RASA that (among other things) will let you connect services to it without needing external tools.
Feel free to jump on our Gitter if you have any questions or need help using it given its (currently) raw state.
Upvotes: 3