search

Xi-yuanWang / GLASS

GLASS: GNN with Labeling Tricks for Subgraph Representation Learning
30 stars 6 forks source link

Query about GNN-seg and GNN-plain #1

Closed shhs29 closed 2 years ago

shhs29 commented 2 years ago

Hi,

I am wondering what is the difference between GNN-seg and GNN-plain mentioned in the paper. I could also not find the implementation of the same in the code.

Thanks a lot in advance, Shweta Ann Jacob

Xi-yuanWang commented 2 years ago

Dear Shweta,

As shown in the ablation analysis section (page 8 in our paper), GNN-plain means GLASS without labeling trick. You can use the following command to run GNN-plain.

python GLASSTest.py --use_nodeid --use_seed --repeat 10 --device $gpu_id --dataset $dataset

GNN-seg is GNN-plain working on the segregated target subgraph rather than the whole graph. I will release the implementation of GNN-seg later.

Sincerely, Xiyuan Wang

shhs29 commented 2 years ago

Hi Xiyuan,

Thanks a lot for the quick response! Looking forward to it.

Closing this issue as it is resolved.

Xi-yuanWang commented 2 years ago

Dear Shweta,

We have released the implementation of GNN-seg. You can use the following command.

python GNNSeg.py --test  --repeat 10 --device $gpu_id --dataset $dataset

Sincerely, Xiyuan Wang

shhs29 commented 2 years ago

Hi Xiyuan,

Thanks a lot for letting me know!