Thank you for responding.
How can I create item2item recommendation data using Movielens data? The model can be like you said. But I am confusing how to make dataset. The thought that comes to mind is to select two from the item numbers (n choose 2) grouped by user and make them related.

Original implicit dastaset (userId, itemId) pair
1,1
1,2
1,3
2,5
2,6
=>
Preprocessed (itemId, itemId) dataset
1,2
2,1
1,3
3,1
2,3
3,2
5,6
6,5