Network science options

It is important to know that you do not need to use ML to work with graphs. ML can just be useful. There is also a blurry line between what is and isn’t ML. For instance, I would consider any form of community detection to be unsupervised ML, as these algorithms are capable of automatically identifying communities that exist in a network. By that definition, we could consider some of the approaches offered by NetworkX unsupervised ML, but they are not given the same level of attention in the data science community, because they are not explicitly called graph ML. There is a level of hype to be aware of.

I am saying this because I want you to keep in mind that there are approaches that you have already learned that can eliminate the need to use what is advertised as graph ML. For instance, you can use Louvain to identify communities, or even just connected components. You can use PageRank to identify hubs – you don’t need embeddings for...