What code can I use to convert training_dict.npy, validation_dict.npy, testing_dict.npy from training_49156.json?

[zozni-douzone]

Thank you very much for your research.

I'm trying to apply this study to my data.
I have a QA dataset in the form of train_49156.json from games, and this code seems to calculate the influence score using the conversion to npy form.

What code can I use to convert training_dict.npy, validation_dict.npy, testing_dict.npy from training_49156.json?

[Linxyhaha]

Thank you very much for your research.

I'm trying to apply this study to my data. I have a QA dataset in the form of train_49156.json from games, and this code seems to calculate the influence score using the conversion to npy form.

What code can I use to convert training_dict.npy, validation_dict.npy, testing_dict.npy from training_49156.json?

Hi,

Thanks for your interest in our work. The npy files store the user interactions in training, validation, and testing sets, which aligns with the data format for the surrogate model, i.e., a sequential recommender model called SASRec, to obtain the influence score.

If you wish to apply the data pruning method to your QA data, the surrogate model used in our work (SASRec) probably doesn't suit your task well. Therefore, you might not necessarily convert the json files into numpy files because they are used for the surrogate sequential recommender model to calculate influence score.

As a suggestion, you might consider utilizing another surrogate model for your specific task (I assume that it is not next-item prediction task) and apply the influence score calculation process to your own surrogate model.