slotalign's Introduction

SLOTAlign

This is a Python implementation of

Robust Attributed Graph Alignment via Joint Structure Learning and Optimal Transport

Jianheng Tang, Weiqi Zhang, Jiajin Li, Kangfei Zhao, Fugee Tsung, Jia Li

ICDE 2023 Arxiv IEEE Library

Dependencies

python 3.9
cuda 11.3
pytorch 1.11
dgl 0.8
pyg 2.0.4
scikit-learn 1.0.2
networkx 2.8.4
argparse 1.4.0

Alignment on Douban Online-Offline

python SLOTAlign.py --config config/douban.json

Alignment on ACM-DBLP

python SLOTAlign.py --config config/dblp.json

Alignment over Inconsistent Structures

python SLOTAlign.py --dataset cora --truncate True --edge_noise 0.5

dataset - cora/citeseer/facebook/ppi
edge_noise - floats between 0 and 1

Alignment over Inconsistent Features

python SLOTAlign.py --dataset cora --noise_type 1 --feat_noise 0.5

dataset - cora/citeseer/facebook/ppi
noise_type - 1: permutation, 2: truncation, 3: compression,
feat_noise - floats between 0 and 1

Alignment on the DBP15K dataset for knowledge graph entity alignment

The dataset and LaBSE embedding files can be downloaded from Google Drive

python run_DBP15K.py

If you use this package and find it useful, please cite our paper using the following BibTeX. Thanks! :)

@inproceedings{tang2023robust,
    author={Tang, Jianheng and Zhang, Weiqi and Li, Jiajin and Zhao, Kangfei and Tsung, Fugee and Li, Jia},
    booktitle = {2023 IEEE 39th International Conference on Data Engineering (ICDE)},
    title = {Robust Attributed Graph Alignment via Joint Structure Learning and Optimal Transport},
    pages = {1638-1651},
    doi = {10.1109/ICDE55515.2023.00129},
    url = {https://doi.ieeecomputersociety.org/10.1109/ICDE55515.2023.00129},
    year={2023}
}

slotalign's People

Contributors

Stargazers

Watchers

slotalign's Issues

a question about the code

Hello, Dear Tang, I am very interested in your work, but I have some problems when reading your code, can you help me to answer it？
When I read the paper, the optimization goal in the paper is shown in the figure below, but when I read your code, the optimization goal in the code is not consistent. How can I explain this?