NeXLink: Node Embedding Framework for Cross-Network Linkages Across Social Networks

This repository contains the code for the implementation of the paper titled NeXLink: Node Embedding Framework for Cross-Network Linkages Across Social Networks by Kaushal et al. published at NetSci-X 2020.

Dataset

The raw graph dataset can be found in /data/raw directory.

Dependencies

The project uses Python 3 dependencies explicitly, for processing and training. All the code is run on JupyterLab computational environment and Anaconda is used as a package manager as well as a virtual environment manager. All the dependencies are exported in the environment.yml file. Make a new environment using:

$ conda env create -f environment.yml

Experiment Steps

• Start by running a Jupyter Lab using $ jupyter lab.
• experiments directory contains construct-graph.ipynb to read the raw dataset and create graph edgelists. Use construct-nodepairs.ipynb to create nodepairs from the graph edgelists. Similarly IONE-dataset.ipynb processes the real world dataset.
• Change the directory to embeddings/OpenNE/src/ and execute run-augmented.sh to extract node embeddings on the augmented dataset and run-ione.sh for the real world dataset. We generate the embeddings using OpenNE open source toolkit.
• Once the embeddings are generated, go back to experiments and run script-retrieval.sh to get results on the augmented dataset and script-retrieval-ione.sh for the real world dataset.
• The results should output in a directory results in the root folder as CSV files.

Citation

If you found this code or our paper useful, please consider citing the following paper:

@inproceedings{kaushal2020nexlink,
    title={NeXLink: Node Embedding Framework for Cross-Network Linkages Across Social Networks},
    author={
        Kaushal, Rishabh and 
        Singh, Shubham and 
        Kumaraguru, Ponnurangam
    },
    booktitle={International Conference on Network Science},
    pages={61--75},
    year={2020},
    organization={Springer}
}