Reviews: Link Prediction Based on Graph Neural Networks

Quality: From the technical point of view, the paper seems to be well-prepared. In particular, the authors propose \gamma-heuristic theory for link prediction and show that many common heuristics can be expressed in a general formulation. They build their framework based on this theory and on the previous work (WLNM). However, I would like to touch on some points here: - In Line 115, there is no clear definition of the \textbf{h-order heuristic} although it is mentioned between lines 26 and 31. If \textbf{h-order heuristic} is defined as, for example, "a heuristic requiring to know up to h-hop neighborhood of the target nodes", then do we really need Theorem 1? - In lines 274 and 275, it is stated that "We run all experiments for 5 times and report the average".