Embedding of Monoplex, Multiplex, Heterogeneous, Multiplex-Heterogeneous and full Multiplex-Heterogeneous Networks.
You can find in this repository the necessary files to use MultiXVERSE for universal multilayer network embedding. You can find the corresponding preprint on osf here: (https://files.osf.io/v1/resources/d78wb/providers/osfstorage/6684cbed9c98c701b51b3f77?action=download&direct&version=3)
To employ MultiXVERSE, it is essential that the networks are formatted similarly to the example found in the /airport directory, as this format is required for utilizing MultiXRank (available at https://github.com/anthbapt/multixrank) that is part of MultiXVERSE. The airport network provided in this directory originates from the same source.
Python 3:
- multixrank
- gensim (fast_version enabled)
- networkx=2.2
- numba=0.50.1
- scikit-learn=0.21.3 (for evaluation)
- pandas
A good way to set up the appropriate environement is to use the associated .yml conda environment.
This program allows to apply MultiXVERSE on a multilayer network, here the airport network.
The usage is the following:
python3 MultiXVERSE.py
The output of this command is the embedding 'embedding_MH.npy' in the directory Results_Embeddings. The embedding is a dictionary with the index as key and the corresponding embedding as value.