Requesting raw files

[SheStat]

Hello! This is really nice and interesting work loved that! However I am trying to regenerate the raw files you used for this study. After using all the filters and matching the mappings I extracted same number of users and items, but the number of interactions is totally different and large. I referred to the previous issues posted by other people regarding this issue as well. But I could not find the answer I need. Is there possibility that you could provide the raw filtered datasets that you used for this study before creating embeddings or profiles.
Thank you.

[Re-bin]

Hi 👋!

Thanks for your interest on RLMRec! Regarding the interaction counts, they should be similar in principle. Here are a few points that might help:

• In our experiments, we applied a filtering criterion to remove interactions with ratings lower than 3 for both the Amazon and Yelp datasets to ensure data quality.
• We used the Amazon 2014 version and the Yelp 2022 version. Different versions may contain varying numbers of interactions.
• Please note that the interaction counts reported in our paper refer only to the training set. If you sum up the interactions across the training, validation, and test sets, the total number would naturally be larger.
• Additionally, it's important to mention that the Yelp and Steam datasets may contain duplicate interactions

I hope the above answer are helpful to you :)

Best regards,
Xubin

[SheStat]

Hello!!

Thank you very much for you kind response, that gives me a solution for a problem I have been struggling for sometime and I am truly thankful. If it is not too much to ask can I know the interactions in the validation and test sets also belong to the same users and items in the training set or are the users and items are also different from the training set. Thank you so much again!

Best Regards,
Shashi.

[Re-bin]

Hi Shashi,

Yes, the set of users and items is the same across training, validation and testing; the only difference is the interactions :)

Best regards,
Xubin

[SheStat]

I sincerely appreciate your prompt response and thank you very much again!!!