TrustAGI-Lab/CoLA

virtual node

Opened this issue · 2 comments

Hello, I would like to ask why there is a need to add an additional virtual node before inputting the node's subgraph and features into the model.

I understand. To conceal the information of the target nodes, and at the same time, create self-loops for the target nodes, we will input them into the Graph Neural Network

I have the same question. When generating subgraphs using rwr, the index of the target node has been added to the walk by subv[i].append(i). So why do we need a new line?