benedekrozemberczki/SimGNN

About dataset

BenedictWongBJUT opened this issue · 6 comments

BenedictWongBJUT commented

I am really interested in this amazing work, but I don't understand how datasets are generated (or processed), or are training data and test data generated from public data sets (just like Linux, AIDS, mentioned in the paper)? I desire to know how syngen.py works and the output of this function.

Thanks a lot.

BenedictWongBJUT commented

I am really interested in this amazing work, but I don't understand how datasets are generated (or processed), or are training data and test data generated from public data sets (just like Linux, AIDS, mentioned in the paper)? I desire to know how syngen.py works and the output of this function.

Thanks a lot.

  • 😄1
lovefxyo commented

I am also curious about this question.

BenedictWongBJUT commented

I am also curious about this question.

I have already figured it out. You can download datasets in this link (https://drive.google.com/drive/folders/1lY3pqpnUAK0H9Tgjyh7tlMVYy0gYPthC?usp=sharing) and with the help of networkX (nx.read_gexf) you can transform the original data (like: 1.gexf) into graph object. And then use syngen.py to get pairwise similarity and generate dataset at the same time. Finally, you can split them into train, validation and test set.

lovefxyo commented

十分感谢！

andWHISKEY commented

I had the same problem. If it's okay with you, can I see the code that uses syngen.py to modify AIDS, etc., other data?

hyzhang-hub commented

Ask if syngen.py is coded by yourself. If so, can you share it