Run rasa nlu in android


(Sitaram Appalla) #1

Hi,

Have few questions -

  1. I would like to run rasa stack in android phone. I know it’s a python lib, and it can run on a local server. Has this been done earlier?
  2. Does it provide sentiment analysis also? If i input “I like Pizza”, and “I don’t like Pizza”, can it differentiate?
  3. Does it work offline without contacting any server?

Sorry for so many questions, and thanks in advance :slight_smile:


(Anders) #2
  1. I guess you can train the language model on a PC and then move the files to Android. But it might be easer to run the bot through an app like Telegram.
  2. Why not try it out? :smiley: I would expect so.
  3. No servers needed! You can run everything from your local machine and rasa is totally GPDR compliant because of that (if you have corporate concerns).

(Sitaram Appalla) #3

Thanks For reply. 1 - Will try out and share results.


(Juste) #4

Keep us updated on how the project goes, @sitaram168. I have played around with kivy a while ago which you can try out if you want to package pre-trained Rasa NLU and Core models inside your app. What I found painful was creating recipes for Rasa NLU and Core libraries so that kivy could use them, but I also certainly didn’t spend enough time working on this project :blush:

Another approach that I played around with was deploying the agent on the cloud (local server would also work) and then sending user inputs as an http request which worked easier than packaging the models inside the app.

Curious to see what you can build! :slight_smile:


(Sitaram Appalla) #5

@Juste Thanks for the tip on Kivy. Will evaluate, and check which one works better :slight_smile: