SWGE

Our implementation of Signed Whole Graph Embeddings methods.

Description

This set of scripts implements the signed whole graph embedding methods presented in our paper. It can be used to:

• Learn the representations of whole signed graphs.
• Perform classification tasks based on the embeddings extracted from signed methods.

Data

We include a few graphs for each dataset in the data folder. The full datasets can be downloaded from [anonymity - available later] Place the downloaded graphs directly into the corresponding folder in data.

Organization

This repository is composed of the following elements:

• Folder data: input signed networks.

• Folder out: contains the files produced by our scripts.

• requirements.py: List of Python packages used in SWGE.

• There are 3 main scripts:

  • SG2V.py: Performs the representation learning step related to the Signed Graph2Vec methods.
  • run_sgcn.py: Performs the representation learning step related to the Signed Graph Convolutional Networks methods.
  • evaluation.py: Performs the classification task.

Installation

• Install Python (tested with Python 3.6.9)
• Install dependencies using the following command:

  pip install -r requirements.txt
  

• In order to use the SGCN method, you have to download the implementation from SGCN and place it inside a SGCN-master folder.
• Retrieve the data from Zenodo and place them into the data folder.

How to use ?

• In order to learn representations with Signed Graph2Vec, run the file SG2V.py. You can configure the model_type between the 3 versions proposed in our paper: g2v, sg2vn or sg2vsb.
• In order to learn representations with SGCN, run the file run_sgcn.py.

These scripts will export the learned representations into the out folder.

• After running the previous scripts, you can perform the classification by running evaluation.py. You can configure the path_emb and path_label variables to change the dataset used.

References

• A. Narayanan, M. Chandramohan, R. Venkatesan, L. Chen, Y. Liu, and S. Jaiswal: graph2vec: Learning distributed representations of graphs, MLG, 2017. URL: [http://www.mlgworkshop.org/2017/paper/MLG2017_paper_21.pdf]
• T. Derr, Y. Ma, and J. Tang: Signed graph convolutional network, 18th ICDM, 2018, DOI: 10.1109/ICDM.2018.00113.