Doesn’t benchmark’s main.py support the implementation of CoLA and RADAR methods?

[uuice11]

No description provided.

[kayzliu]

Thanks for your issue. Yes, Radar is included in in our benchmark, but CoLA is not. In the original paper of CoLA, random walk with restart is used to sample the subgraph. However, we implement CoLA by random neighbor sampling, which may affect the performance (can be better and can be worse). Thus, our implementation of CoLA has not been included in the benchmark.