Fix GCMC Invalid Key Error

[darrylong]

Description

For instances when predicting scores and item_idx is none:

• It currently throws a "invalid key" error for something that has not been seen

Solved by returning a default score instead.

Related Issues

Checklist:

• I have added tests.
• I have updated the documentation accordingly.
• I have updated README.md (if you are adding a new model).
• I have updated examples/README.md (if you are adding a new example).
• I have updated datasets/README.md (if you are adding a new dataset).

[tqtg]

@darrylong to make it right, I think we need to do something similar to the predict function. Basically, we need to create a test decoding graph for that particular user (connecting from the user node to all item nodes), then call the model and obtain predicted ratings for all items. Let me know what you think.

[darrylong]

@darrylong to make it right, I think we need to do something similar to the predict function. Basically, we need to create a test decoding graph for that particular user (connecting from the user node to all item nodes), then call the model and obtain predicted ratings for all items. Let me know what you think.

I think that should work. I'll give it a try!

[darrylong]

@tqtg while I was attempting to add individual dec_graphs of each user, I thought about the possibility of adding all items for users in the test set into the dec_graph. This could potentially help with the retrieval of scores, but would like to check if this is theoretically correct.

[tqtg]

Do you mean we compute scores for all users in the test set at once?

[darrylong]

Do you mean we compute scores for all users in the test set at once?

That's right. In a single dec_graph just like in the predict function.