This is the source code for Unsupervised Social Networks Alignment via Cross Network Embedding.

• initial_data_processing.py & initial_data_processing_MC3.py: Transforming the raw data into a suitable format for code to run.
• main.py : The main function to run the code.

python initial_data_processing.py -n 5000 -g 5 -r 0.05 -d lastfm

python initial_data_processing_MC3.py

python main.py -d flickr-flickr

• Python == 3.5.2
• NumPy
• PyTorch == 1.4.0
• networkx

We used four public datasets in the paper:

• Last.fm
• Flickr
• MySpace

You can learn more about these three datasets from here.

• MC3

You can get the processed datasets in -data/data_backup.

If you want to use the back up data, please move the corresponding data to -data/graph_edge.

If you want to try the experiment across networks(Last.fm, Flickr and MySpace), we provide the processed real user mapping in -data/graph_map.