mOWL is a software library that incorporates several methods to generate embeddings of entities in ontologies. This project started during the BioHackathon Europe edition in 2021, where we made significant progress in setting the bases of the library, developing some methods, creating datasets, and starting the documentation. For this year, we propose the continuation of the development of mOWL. The current state of mOWL contains several methods categorized into graph-based, syntactic, and semantic. Furthermore, we provide some datasets related to protein-protein interactions as well as Jupyter notebook tutorials. This project can be continued and extended by adding more methods, optimizing the existing ones and creating other functionalities such as a standardized evaluation framework. Additionally, documentation for the library can also be improved. We expect to have a fully working version that we can publish in the main Python package repositories such as PyPi and Conda. The project is available at https://github.com/bio-ontology-research-group/mowl. A testing version of the library is available at https://test.pypi.org/project/mowl-borg/ and current documentation is available at https://mowl.readthedocs.io/en/latest/index.html
Interoperability Platform Machine learning Tools Platform
Project Number: 18
Maxat Kulmanov, [email protected]
A library and toolkit, together with a set of biomedical use cases/examples and documentation. It is expected to be done in 4 days.
Participants can provide use cases, implement algorithms, design new algorithms, test the library, and provide documentation and tutorials. Skills needed: • machine learning • Python, Java or Scala programming • ontologies, Web Ontology Language (OWL) • reasoning • knowledge graphs
Number of expected hacking days: 4 days