REGCN

This repo provides a reference implementation of REGCN as described in the paper:

Yang, J., Zhou, W., Wei, L., Lin, J., Han, J., Hu, S. (2020). RE-GCN: Relation Enhanced Graph Convolutional Network for Entity Alignment in Heterogeneous Knowledge Graphs. In: Nah, Y., Cui, B., Lee, SW., Yu, J.X., Moon, YS., Whang, S.E. (eds) Database Systems for Advanced Applications. DASFAA 2020. Lecture Notes in Computer Science(), vol 12113. Springer, Cham. https://doi.org/10.1007/978-3-030-59416-9_26

Basic Usage

Code reference

• Config.py：实验参数配置。
• evaluation.py：提供测试中的距离、准确率计算功能。
• IOUtils.py：文件读写模块。
• layers.py：模块化GCN。
• model.py：REGCN模型。
• pp_JLER.py：提供数据预处理功能，包含数据集分割、训练、测试。
• run.py：实验入口程序。
• utils.py：工具箱，多个功能实现。文件未处理，可能包含了多个其他实验过程所需代码。

Run the code

由于这是后期其他同学帮忙整理出的代码，代码虽然可以跑通，但是不能保证实验结果完全相同。

运行前需要将数据与代码组织成如下结构：

cd ./REGCN

# run the model 
python run.py 

Datasets

the datasets could be found in the following links: REGCN_data

Cite

If you find REGCN useful for your research, please consider citing us :

@InProceedings{10.1007/978-3-030-59416-9_26,
    author="Yang, Jinzhu
    and Zhou, Wei
    and Wei, Lingwei
    and Lin, Junyu
    and Han, Jizhong
    and Hu, Songlin",
    editor="Nah, Yunmook
    and Cui, Bin
    and Lee, Sang-Won
    and Yu, Jeffrey Xu
    and Moon, Yang-Sae
    and Whang, Steven Euijong",
    title="RE-GCN: Relation Enhanced Graph Convolutional Network for Entity Alignment in Heterogeneous Knowledge Graphs",
    booktitle="Database Systems for Advanced Applications",
    year="2020",
    publisher="Springer International Publishing",
    address="Cham",
    pages="432--447",
    isbn="978-3-030-59416-9"
}