map the nodes to embeddings about verse

[xgfs]

Note that I still sort the nodes, so the order should be deterministic.

[hadisfr]

convert script generates a node2id map to convert IDs to sequential integers which is used in BCSR. You can change the converter script to save this mapping somewhere.

verse/python/convert.py

Line 67 in 51fcfe9

node2id = dict(zip(sorted(map(int, nodes)), range(number_of_nodes)))

Note that because this mapping is generated based on a set of nodes, there is no guaranty on ordering of nodes, so the generated mapping may be vary each time you run the convert script.

[hadisfr]

Note that I still sort the nodes, so the order should be deterministic.

Oh yes. I missed the sort part. 🤦🏻‍♂️

[linz2000]

Hi Author,
I also want to know how to map the node to its embedding, and I am confused about using the same train data but getting different results each time?

[xgfs]

What do you mean by results? By default, the embeddings are initalized randomly each time, so it's normal that you get different embedding each time you run the model.

[linz2000]

oh yes, I mean got different embeddings. Could I ask how to get the embedding of each node? Are these embedding results sorted in ascending order by node number?

[xgfs]

If you converted the graphs with the supplied converter, then the node IDs are sorted in the ascending order.

[linz2000]

I got it, thanks for your reply!