The idea of combining community detection and learning node representation is quite natural but apparently it is not well entrenched in the network analysis community. The reviewers feel that for this reason the paper would be a valuable contribution to the field.