CaptumExplainer subgraph results interpretation

[mariakritharidou]

Hello everyone. I’m using CaptumExplainer to generate some explanations for my heterogenous graph for binary classification link prediction. I pass the edge I want to produce the explanations for (that is a non existent edge in the graph) and by using the get_explanation_subgraph method I get the subgraph. When viewing the results, let’s assume the link for prediction was nodes 5 to 17, I don’t see in the results those nodes. I would expect the subgraph to include those nodes and the neighbours around them. Do I do something wrong?