This open-source project contains the Python implementation of our approach Shallom, training and evaluation scripts. We added Shallom into Knowledge Graph Embeddings at Scale open-source project to ease the deployment and the distributed computing. Therein, we provided pre-trained models on many large knowledge graphs.
Knowledge graph completion refers to predicting missing triples. Most approaches achieve this goal by predicting entities, given an entity and a relation. We predict missing triples via the relation prediction. To this end, we frame the relation prediction problem as a multi-label classification problem and propose a shallow neural model (SHALLOM) that accurately infers missing relations from entities. SHALLOM is analogous to C-BOW as both approaches predict a central token (p) given surrounding tokens ((s,o)). By virtue of its architecture, SHALLOM requires a maximum training time of 8 minutes on benchmark datasets including WN18RR, FB15K-237 and YAGO3-10. Hence, one does not need to win the hardware lottery to use SHALLOM for predicting missing information on knowledge graphs.
- DBpedia embeddings
- Carcinogenesis embeddings
- Mutagenesis embeddings
- Biopax embeddings
- Family embeddings
First clone the repository:
git clone https://github.com/dice-group/Shallom.git.
Then obtain the required libraries:
conda env create -f environment.yml
source activate shallom
The code is compatible with Python 3.6.4.
- To reproduce the reported results for our approach, please refer to the any desired .ipynb file.
- Run any desired .ipynb file
If you use SHALLOM, please cite the following publication:
@inproceedings{demir2021shallow,
title={A shallow neural model for relation prediction},
author={Demir, Caglar and Moussallem, Diego and Ngomo, Axel-Cyrille Ngonga},
booktitle={2021 IEEE 15th International Conference on Semantic Computing (ICSC)},
pages={179--182},
year={2021},
organization={IEEE}
}
For any further questions, please contact: caglar.demir@upb.de
or caglardemir8@gmail.com
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.