[mo_ibrahem]

Hello Everyone,
I‘m Mostafa Ibrahem,4th year Computer Science student at Ain Shams University From Cairo,Egypt.I’m interested in “Using machine learning to improve suggestion”.By having a nice experience in working in recommendation engines during my research project at university of Nebraska at Lincoln USA.And my Graduation Project this year is mainly a hybrid recommendation engine .

For preparing my draft proposal for Kodi and mainly the proposal will introduce the hybrid approach recommendation system “mix of content based and collaborative filtering based”
I’m willing to use Python-TensorFlow  as I certified Machine Learning with TensorFlow on Google Cloud Platform

Valuable suggestions to improve the preparation of the proposal is much much welcome 

Best regards,

[Razze]

How would Kodi consume those recommendations according to your vision?

[mo_ibrahem]

How would Kodi consume those recommendations according to your vision?

It should be optimally an API running on sand-alone  server  to not affect the core code (of-course it shouldn't affect the core code) after deployment the model will receive the feedback by the API from the home page of the application we can depend on two factors :
user preference->user rating /like
similar user activity for ex if we have two users A and B :if A likes rap song and jazz music  same as B 
after that lets's say user A like country song it may be recommended to user B (as the same 2 users have similar behavior we can see that user B got recommendation without even express any interest in country music ) 
as same regards to movies