For some years now, the scientific community has come to recognize the need of sharing not only final methods and results via scholarly publications but also complementary research objects (ROs) such as software and data. To maximize their usefulness, ROs should be accompanied by metadata, interconnected to each other in meaningful ways, and made as open as possible (and as close as necessary). Linked Open Science (i.e., Open Science plus Linked Open Data extended to all sorts of ROs) deals with the effective combination across these three dimensions to improve, for instance, transparency and reproducibility in science. FAIR and openness efforts around data have already improved the situation but more needs to be done for other ROs such as software, workflows, machine learning, grants, management plans, etc. At this BioHackathon, we aim at defining metadata schemas under the Bioschemas umbrella, for not yet covered ROS. In particular, we plan to work on metadata for management plans and machine learning but we are open to any other ROs that participants are interested in. We will base our metadata schemas on previous work done by two focus groups part of the ELIXIR Tools Platform, viz. the Good Practices group which provides a questionnaire for Software Management Plans, and the Machine Learning group which provides the DOME recommendations for supervised learning. We will also include work done by groups in the Research Data Alliance viz. the Data Management Plans Common Standards Working Group and the FAIR for Machine Learning Interest Group, as well as analyses on machine learning metadata carried out by the German national project NFDI4DataScience.
Bioschemas Interoperability Platform Machine learning Tools Platform
Project Number: 17
- Leyla Jael Castro ([email protected])
- Dietrich Rebholz-Schuhmann ([email protected])
- Ivan Micetic ([email protected])
- Metadata assessment on some platforms supporting the selected research objects (e.g., related to machine learning or research management plans)
- Metadata crosswalk for selected research objects
- Draft definition of Bioschemas profiles for selected research objects
- People with some knowledge on metadata schemas
- People with some knowledge on Data/Software Management Plans
- People with some knowledge on parameters and good practices for Machine Learning
- People interested in research metadata
Number of expected hacking days: 4