This is a simple Jupyter notebook with a demo implementation of the "PENELOPIE: Enabling Open Information Extraction for the Greek Language through Machine Translation" paper, to be presented in EACL 2021 SRW.
- Python 3.6.9
- The Python wrapper of OpenIE5 (has to be installed manually)
- The Python wrapper of ClausIE (has to be installed from source)
- Follow the instructions below for additional requirements.
- It is recommended to create a new virtual environment to avoid version conflicts. For Linux:
python3 -m venv penelopie
source penelopie/bin/activate- Install the OpenIE5 and ClausIE OIE engines following the above links.
- Install additional requirements with:
git clone https://github.com/lighteternal/PENELOPIE.git
cd PENELOPIE
pip install -r requirements.txtThe pipeline showcased in the Jupyter notebook will process the Greek sentences in data/demo_greek.txt, translate them into English, extract triples using 3 OIE engines and back-translate the resulting triples in Greek. The results are found in data/final_triples.csv.
The standalone NMT models (EN2EL, EL2EN) can be found here: https://huggingface.co/lighteternal
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.