search

apple / turicreate

Turi Create simplifies the development of custom machine learning models.
BSD 3-Clause "New" or "Revised" License
11.2k stars 1.14k forks source link

Sorry, I want to know the specific algorithm of item similarity recommender #559

Closed tjnu1430130310 closed 6 years ago

tjnu1430130310 commented 6 years ago

Hello, I just started learning machine, English is not my mother tongue, I'm very sorry.

I built a recommendation system using GraphLab Create and Django. Now I really want to know whether the algorithm behind the use of item_similarity_recommender modeling is collaborative filtering. In addition, can you tell me more about whether it is user-based collaborative filtering or item-based collaborative filtering?

I know I shouldn't ask this question to you. I'm very sorry, but I'm always puzzled and I'm so sorry.

tjnu1430130310 commented 6 years ago

In here, https://apple.github.io/turicreate/docs/api/generated/turicreate.recommender.item_similarity_recommender.create.html#turicreate.recommender.item_similarity_recommender.create

Create a recommender that uses item-item similarities based on users in common.

Does this sentence mean that it is the user-based collaborative recommendation?

tjnu1430130310 commented 6 years ago

I noticed there was an itemCF when I see the code, so is this a item-based collaborative filtering?

boulbi777 commented 5 years ago

Hello, why did you close this issue ? I'm really interested since it's been confusing for me. Have you found a solution ? If so I'd be glad to have a response ! Thank you ! 👍

nickjong commented 5 years ago

Yes, this toolkit is an implementation of collaborative filtering. For some more details, try looking at the relevant C++ header: https://github.com/apple/turicreate/blob/master/src/unity/toolkits/recsys/models/itemcf.hpp