The movie id (both in column "Movie ID" and "user_hist") starts from 1 (Evidence: line 162 needs cur_id-1 to get the corresponding row), while the ids in cur_indice starts from 0. When considering whether the current movie id (from cur_indice) in the input_dict["user_hist"] (line 172), the index should plus 1 or the movie ids in user history should all minus 1. I find that the corresponding parts of the other two datasets has changed the way to selecte the most relevant items, which I haven't checked yet. Looking forward to your reply.
Dear authors,
Thanks for your excellent work!
However, I want to post an item-id mismatch problem, which may lead to retrieve wrong relevant items (Maybe I'm wrong). Here's the problem:
The movie id (both in column "Movie ID" and "user_hist") starts from 1 (Evidence: line 162 needs
cur_id-1to get the corresponding row), while the ids incur_indicestarts from 0. When considering whether the current movie id (fromcur_indice) in theinput_dict["user_hist"](line 172), theindexshould plus 1 or the movie ids in user history should all minus 1. I find that the corresponding parts of the other two datasets has changed the way to selecte the most relevant items, which I haven't checked yet. Looking forward to your reply.