Edge Classification on the Citations Dataset

[Qfl3x]

Hello, I'm looking for a solution to the problem of Edge Classification though having trouble with the definition of the problem. What I thought I would pass as input two feature vectors and the output to be the probability of these two papers being linked (non-abelian process). But I'm struggling on how to pass that onto the dataset as it takes x as the node-features.