dwjpman / GGAE

GGAE: Global Graph Attention Embedding Network for Relation Prediction in Knowledge Graphs

Source code for our [TNNLS] paper: [Global Graph Attention Embedding Network for Relation Prediction in Knowledge Graphs]

Requirements

• conda

Please download miniconda from above link and create an environment using the following command:

    conda env create -f pytorch35.yml

Activate the environment before executing the program as follows:

    source activate pytorch35

Dataset

We used three different datasets for evaluating our model. All the datasets and their folder names are given below.

• Freebase: FB15k-237
• Wordnet: WN18RR
• Kinship: kinship

Reproducing results

To reproduce the results published in the paper:

    $ bash train_xx.sh

Citation

Please cite our paper if you use this code in your work.

    @ARTICLE{9452178,
      author={Li, Qian and Wang, Daling and Feng, Shi and Niu, Cheng and Zhang, Yifei},
      journal={IEEE Transactions on Neural Networks and Learning Systems}, 
      title={Global Graph Attention Embedding Network for Relation Prediction in Knowledge Graphs}, 
      year={2021},
      volume={},
      number={},
      pages={1-14},
      doi={10.1109/TNNLS.2021.3083259}}

For any clarification, comments, or suggestions please create an issue or contact [e-mail]