Origin of the datasets about selfrec HOT 4 CLOSED

Furthermore, I'm not able to train the sequential models such as SASRec and CL4SRec with other datasets such as amazon-beauty. I see this one has a different format compared to amazon-kindle and yelp2018. Could you advice on the difference between these data processing? And how you ran all the models on yelp2018 to reproduce the SOTs?

from selfrec.

link of amazon-book: https://www.dropbox.com/sh/20l0xdjuw0b3lo8/AABBZbRg9hHiN42EHqBSvLpta?dl=0

I got this dataset from https://github.com/huangtinglin/NGCF-PyTorch/tree/master/Data

format of datasets for sequential models:
user: interacted items

from selfrec.

Hello - Thank you for your response.
But the dataset don't seem to be the same format for amazon-book dataset. For instance for test.txt what are these columns?

4 0 1
7 0 1
9 0 1
17 2 1
19 2 1
21 3 1
29 4 1
31 4 1

And for the Yelp2018, how do you process the data into the user: interacted items format? Because only amazon-beauty has the correct form to be trained with a sequential model.

from selfrec.

1 means interacted. You can regard it as a placeholder.

Yelp2018 is processed for graph recommenders, not for sequential models.

from selfrec.