William L Hamilton

Hi, in this case we can use one-hot indicator vectors as the features for each node. I.e., we can assign node $i$ as a vector of size |V| where all...

One thing that is necessary is to constrain the embeddings to be unit length. This is mentioned in the appendix of the paper I think. For instance you can use...

Sadly, I don't plan on implementing the unsupervised any time soon, but pull requests are welcome! :)

Hi, So this is somewhat unfortunate, but if I recall correctly that "{}-dict.pkl" is actually a vocabulary object from gensim's word2vec object (which no longer exists, since gensim did major...

This is not a priority; not sure if it will make that much of a difference...

Yes.. this is just an oversight in the way we set things up. It is not really needed, and we hope to remove the unnecessary dependency soon.

Hi @doannam020293, Thanks for the question! Could you describe your use case/constraints in a bit more detail? Right now, we simulate adding new nodes after training by marking some nodes...

Numpy can't natively handle or distribute large matrix computations that are needed. I think the solution is to write some cython/c code to handle the Dinv.dot(L).dot(Dinv) computation.

Hi, Yes currently, there the code is only compatible with Keras 0.3 and not the newer versions. I am hoping to fix this, but I am quite busy with other...