Discover KRR-Oxford/DeepOnto Open Source project by KRR (@KRR-Oxford)

A package for ontology engineering with deep learning.

News 📰

Check the complete changelog and FAQs. The FAQs page does not contain much information now but will be updated according to feedback.

About

$\textsf{DeepOnto}$ aims to provide building blocks for implementing deep learning models, constructing resources, and conducting evaluation for various ontology engineering purposes.

Documentation: https://krr-oxford.github.io/DeepOnto/.
Github Repository: https://github.com/KRR-Oxford/DeepOnto.
PyPI: https://pypi.org/project/deeponto/.

Installation

OWLAPI

$\textsf{DeepOnto}$ relies on OWLAPI version 4 (written in Java) for ontologies.

We follow what has been implemented in mOWL that uses JPype to bridge Python and Java Virtual Machine (JVM). Please check JPype's installation page for successful JVM initialisation.

Pytorch

$\textsf{DeepOnto}$ relies on Pytorch for deep learning framework.

We recommend installing Pytorch prior to installing DeepOnto following the commands listed on the Pytorch webpage. Notice that users can choose either GPU (with CUDA) or CPU version of Pytorch.

In case the most recent Pytorch version causes any incompatibility issues, use the following command (with CUDA 11.6) known to work:

pip3 install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116

Basic usage of DeepOnto does not rely on GPUs, but for efficient deep learning model training, please make sure torch.cuda.is_available() returns True.

Install from PyPI

Other dependencies are specified in setup.cfg and requirements.txt which are supposed to be installed along with deeponto.

# requiring Python>=3.8
pip install deeponto

Install from Git Repository

To install the latest, probably unreleased version of deeponto, you can directly install from the repository.

pip install git+https://github.com/KRR-Oxford/DeepOnto.git

Main Features

Figure: Illustration of DeepOnto's architecture.

Ontology Processing

The base class of $\textsf{DeepOnto}$ is [Ontology][deeponto.onto.Ontology], which serves as the main entry point for introducing the OWLAPI's features, such as accessing ontology entities, querying for ancestor/descendent (and parent/child) concepts, deleting entities, modifying axioms, and retrieving annotations. See quick usage at load an ontology. Along with these basic functionalities, several essential sub-modules are built to enhance the core module, including the following:

Ontology Reasoning ([OntologyReasoner][deeponto.onto.OntologyReasoner]): Each instance of $\textsf{DeepOnto}$ has a reasoner as its attribute. It is used for conducting reasoning activities, such as obtaining inferred subsumers and subsumees, as well as checking entailment and consistency.
Ontology Pruning ([OntologyPruner][deeponto.onto.OntologyPruner]): This sub-module aims to incorporate pruning algorithms for extracting a sub-ontology from an input ontology. We currently implement the one proposed in [2], which introduces subsumption axioms between the asserted (atomic or complex) parents and children of the class targeted for removal.
Ontology Verbalisation ([OntologyVerbaliser][deeponto.onto.OntologyVerbaliser]): The recursive concept verbaliser proposed in [4] is implemented here, which can automatically transform a complex logical expression into a textual sentence based on entity names or labels available in the ontology. See verbalising ontology concepts.
Ontology Projection ([OntologyProjector][deeponto.onto.OntologyProjector]): The projection algorithm adopted in the OWL2Vec* ontology embeddings is implemented here, which is to transform an ontology's TBox into a set of RDF triples. The relevant code is modified from the mOWL library.
Ontology Normalisation ([OntologyNormaliser][deeponto.onto.OntologyNormaliser]): The implemented $\mathcal{EL}$ normalisation is also modified from the mOWL library, which is used to transform TBox axioms into normalised forms to support, e.g., geometric ontology embeddings.
Ontology Taxonomy ([OntologyTaxonomy][deeponto.onto.OntologyTaxonomy]): The taxonomy extracted from an ontology is a directed acyclic graph for the subsumption hierarchy, which is often used to support graph-based deep learning applications.

Tools and Resources

Individual tools and resources are implemented based on the core ontology processing module. Currently, $\textsf{DeepOnto}$ supports the following:

BERTMap [1] is a BERT-based ontology matching (OM) system originally developed in repo but is now maintained in $\textsf{DeepOnto}$. See Ontology Matching with BERTMap & BERTMapLt.
Bio-ML [2] is an OM resource that has been used in the Bio-ML track of the OAEI. See Bio-ML: A Comprehensive Documentation.
BERTSubs [3] is a system for ontology subsumption prediction. We have transformed its original experimental code into this project. See Subsumption Inference with BERTSubs.
OntoLAMA [4] is an evaluation of language model probing datasets for ontology subsumption inference. See OntoLAMA: Dataset Overview & Usage Guide for the use of the datasets and the prompt-based probing approach.

License

!!! license "License"

Copyright 2021-2023 Yuan He.
Copyright 2023 Yuan He, Jiaoyan Chen.
All rights reserved.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at *<http://www.apache.org/licenses/LICENSE-2.0>*

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.

Citation

The preprint of our system paper for $\textsf{DeepOnto}$ is currently available at arxiv.

Yuan He, Jiaoyan Chen, Hang Dong, Ian Horrocks, Carlo Allocca, Taehun Kim, and Brahmananda Sapkota. DeepOnto: A Python Package for Ontology Engineering with Deep Learning. arXiv preprint arXiv:2307.03067 (2023).

@article{he2023deeponto,
  title={DeepOnto: A Python Package for Ontology Engineering with Deep Learning},
  author={He, Yuan and Chen, Jiaoyan and Dong, Hang and Horrocks, Ian and Allocca, Carlo and Kim, Taehun and Sapkota, Brahmananda},
  journal={arXiv preprint arXiv:2307.03067},
  year={2023}
}

Relevant Publications

[1] Yuan He‚ Jiaoyan Chen‚ Denvar Antonyrajah and Ian Horrocks. BERTMap: A BERT−Based Ontology Alignment System. In Proceedings of 36th AAAI Conference on Artificial Intelligence (AAAI-2022). /arxiv/ /aaai/
[2] Yuan He‚ Jiaoyan Chen‚ Hang Dong, Ernesto Jiménez-Ruiz, Ali Hadian and Ian Horrocks. Machine Learning-Friendly Biomedical Datasets for Equivalence and Subsumption Ontology Matching. The 21st International Semantic Web Conference (ISWC-2022, Best Resource Paper Candidate). /arxiv/ /iswc/
[3] Jiaoyan Chen, Yuan He, Yuxia Geng, Ernesto Jiménez-Ruiz, Hang Dong and Ian Horrocks. Contextual Semantic Embeddings for Ontology Subsumption Prediction. World Wide Web Journal （WWWJ-2023). /arxiv/ /wwwj/
[4] Yuan He‚ Jiaoyan Chen, Ernesto Jiménez-Ruiz, Hang Dong and Ian Horrocks. Language Model Analysis for Ontology Subsumption Inference. Findings of the Association for Computational Linguistics (ACL-2023). /arxiv/ /acl/
[5] Yuan He, Jiaoyan Chen, Hang Dong, and Ian Horrocks. Exploring Large Language Models for Ontology Alignment. ISWC 2023 Posters and Demos: 22nd International Semantic Web Conference. /arxiv/ /iswc/

Please report any bugs or queries by raising a GitHub issue or sending emails to the maintainers (Yuan He or Jiaoyan Chen) through:

[email protected]

KRR-Oxford/DeepOnto

KRR-Oxford

Reviews

Repository Details