File name sort is incorrect.

[Manslow]

Thanks for the fantastic embedding approach and implementation! I'm currently using it for a project I'm working on.

I found a possible bug in your code. In the get_files function of the utils module you have the following lines:

all_files = list(set(all_files))
all_files.sort()

I'm assuming that you want to sort graph files in order from low ids to high ids. However, .sort() will try to the sort the file name strings based on the bytes that make up the string not the int representation of it. In my work this led to an incorrect mapping between embeddings and graphs as I assumed the order would be the same.

I changed this to the rather ugly:

all_files = sorted(all_files, key=lambda file_name: int(file_name.split('/')[-1].split('.')[0]))

which extracts the file name then the graph number and sorts it as an int.

I'm not sure if this makes a difference for your paper but it might be worth reviewing.