Request for the Yelp2018 Dataset Used in the Paper or Guidance on Preparing a Custom Dataset #2
Description
hello，
I’ve been studying your paper on the AdvDrop model, and I find it highly insightful and valuable for advancing research in this area. I’m particularly interested in the model’s potential and have been working to replicate your results.
However, while examining the code, I noticed that the training dataset yelp2018.new contains 937,416 interactions, while the paper mentions a total of 134,031 interactions in the Yelp2018 dataset. This discrepancy raised a few questions on my end, and I was hoping to kindly ask if you could provide the original Yelp2018 dataset that was used in the paper.
If that dataset cannot be shared, would you be able to offer guidance on how to preprocess and split a custom dataset to match the format used by your code? Specifically, I am interested in knowing how to properly structure the data so that it is compatible with the model, as well as the ideal way to partition the dataset for training, evaluation and test.
Once again, I want to express my sincere admiration for your work on the AdvDrop model. It has inspired me to dive deeper into this area, and I look forward to any guidance you might offer.
Thank you very much for your time and help!