deepsnap
deepsnap copied to clipboard
How to generate WN18.gpickle or How to format own KGs
I created a KG and intended to split them into train/val/test without negative triples. I saved my KG as a multidigraph (networkx) and set resample_negatives=False. However, the splitting graphs still contain many non-existing edges. But the data "WN18.gpickle" you provided did not have this issue. Could you please tell me how you generate the WN18.gpickle? Thank you very much for your help.