HI there I want to ask the way to produce my own dataset

[johnnyjana730]

I find that the dataset you use there is in *.pkl format. I am a little confused how to produce my own dataset. I read the code in "train.py" and it seem like in the pkl file has three key 'train_data', 'test_data', 'adj'.

I want to ask whether these three value is exactly same to the dataset in original RippleNet repositories dataset, which is in *.txt format. In other word, 'train_data' and 'test_data' split from "ratings_final.txt" and 'adj' is same to "kg_final.txt"? If is it right it mean that if I want to produce my own dataset I just need to sort my own dataset same to RippleNet's one and transfer the data of "ratings_final.txt" and "kg_final.txt" to 'train_data', 'test_data', 'adj' and save it in pkl file?

[Atomu2014]

Hi, sorry for the late response... I missed the notification... The data format is not the same as RippleNet, and the processing code can be found under "process/". BTW, you can download my processed dataset and check the pickle files.