Skip to content
SEAL (learning from Subgraphs, Embeddings, and Attributes for Link prediction). "M. Zhang, Y. Chen, Link Prediction Based on Graph Neural Networks, NIPS 2018 spotlight".
Branch: master
Clone or download
Latest commit fcb7d57 Apr 30, 2019
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
MATLAB reset Main.m to only basic experiments Oct 20, 2018
Python fix a bug from nx.from_scipy_sparse_matrix that treats target link (0… Mar 26, 2019
README.md readme update Apr 30, 2019

README.md

SEAL -- learning from Subgraphs, Embeddings, and Attributes for Link prediction

About

Code for SEAL (learning from Subgraphs, Embeddings, and Attributes for Link prediction). SEAL is a novel framework for link prediction which systematically transforms link prediction to a subgraph classification problem. For each target link, SEAL extracts its h-hop enclosing subgraph A and builds its node information matrix X (containing latent embeddings and explicit attributes of nodes). Then, SEAL feeds (A, X) into a graph neural network (GNN) to classify the link existence, so that it can learn from both graph structure features (from A) and latent/explicit features (from X) simultaneously for link prediction.

For more information, please check our paper:

M. Zhang and Y. Chen, Link Prediction Based on Graph Neural Networks, Advances in Neural Information Processing Systems (NIPS-18). [PDF]

Version

SEAL is implemented in both MATLAB and Python. The MATLAB version was used to generate the experimental results in the paper, which also contains the evaluation code of other baseline methods. The Python software has better flexibility and scalability.

Reference

If you find the code useful, please cite our paper:

@inproceedings{zhang2018link,
  title={Link prediction based on graph neural networks},
  author={Zhang, Muhan and Chen, Yixin},
  booktitle={Advances in Neural Information Processing Systems},
  pages={5165--5175},
  year={2018}
}

Muhan Zhang, Washington University in St. Louis muhan@wustl.edu 9/5/2018

You can’t perform that action at this time.