dbbice / DMGNN

Overview

We use Python and Pytorch to implement a Disentangled Multi-view Graph Neural Network for Multilingual Knowledge Graph Completion model named DMGNN.

Getting Started

Datasets

We use [DBP-5L] and [E-PKG] datasets. The structure of DBP-5L dataset is listed as follows:

datasetdbp5l/:
├── entities/
│   ├── el.tsv: entity names for language 'el'
├── kg/
│   ├── el-train.tsv: the train dataset for the completion task
│   ├── el-val.tsv: the train dataset for the completion task
│   ├── el-test.tsv: the train dataset for the completion task
├── seed_train_pairs/
│   ├── el-en.tsv: alignment training seeds
├── seed_train_pairs/
│   ├── el-en.tsv: alignment test seeds
├── relation.txt: set of relations

install dependencies

pip install -r requirements.txt

## Experiments
### Training and Testing

To reproduce our experiments, please use the following script:

```bash
# w/ SI
python train.py --data_path datasetdbp5l/ --target_language ja
# w/o SI
python train.py --data_path datasetdbp5l/ --target_language ja --no_name_info --dropout 0.1