Is there a way to apply GCN to directed graphs？

[XJTUAlan]

I've only recently started working on GCN, and I've found that it usually works on undirected graphs. But, I have a question: is there any way to apply GCN to directed graphs?

[tkipf]

Yes! See https://arxiv.org/abs/1703.06103

…

On Thu, Jun 25, 2020 at 14:12 Alan Yang ***@***.***> wrote: I've only recently started working on GCN, and I've found that it usually works on undirected graphs. But, I have a question: is there any way to apply GCN to directed graphs? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#175>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABYBYYGZMSDBAMB2VUKIGG3RYM5JNANCNFSM4OIJYNPQ> .

[kepsail]

https://arxiv.org/abs/2005.13183

[QasimWani]

wouldn't converting to bipartite graph as the paper suggested in the end work just as well?
in reference to https://www.microsoft.com/en-us/research/wp-content/uploads/2017/01/SDG.pdf