swapagarwal / JARVIS-on-Messenger

:speech_balloon: A community-driven python bot that aims to be as simple as possible to serve humans with their everyday tasks
https://m.me/J.A.R.V.I.S.on.Messenger
MIT License
1.35k stars 990 forks source link

Support for AIML2 - training files #27

Closed johndpope closed 8 years ago

johndpope commented 8 years ago

https://github.com/creatorrr/pyAIML

swapagarwal commented 8 years ago

Can you give some example(s) showing how can we utilize this?

johndpope commented 8 years ago

https://github.com/pandorabots/rosie/tree/master/lib/aiml

KartikKannapur commented 8 years ago

@johndpope Thanks for sharing.

AIML seems to take a rule-based approach in building a question-answering system, where most of the responses would have to be hard coded. I think JARVIS would not scale with such a limitation. Also, most of the projects/libraries around AIML are not currently maintained. I suppose we could stick to a text-mining based approach rather than a hard coded rule-based one.

@swapagarwal Your thoughts?

swapagarwal commented 8 years ago

Hi @johndpope, I took a look at AIML and I totally agree with @KartikKannapur. JARVIS aims to learn from human language and respond accordingly. Using AIML will restrict it to a fixed set of queries and scaling it would be quite a tedious task.

johndpope commented 8 years ago

I will close this ticket - fyi

screen shot 2016-05-02 at 7 13 21 pm

https://docs.api.ai/docs/concept-entities#system-entities

https://docs.api.ai/v15/docs/apiai-in-2-minutes-1

swapagarwal commented 8 years ago

@johndpope we're using an api.ai alternative, see https://wit.ai/swapagarwal/JARVIS-on-Messenger/