acbull
commented
2 years ago Hi.
The sample_subgraph is a layer-wise algorithm to sample a subgraph (which we describe in the paper), and inp is the input feature.
In this github repo I mainly keep my own data structure, but if you want to do your own research work, I might recommend you to laod the data (especially large-scale heterogeneous graph) using DGL or PyG's data structure and sampling API, they should more clean to use.
Dear Author, I am trying to apply the HGT model in my research work. For that I am doing preprocess work, I found your data.py and Preprocess_OAG.py very helpful. However, I could not able to understand the "inp and sampled depth" defined in the data.py file under the pyHGT folder. I am looking forward for your reply