osparc-classifiers

A curated list of classifiers defined and used by the osparc-community

Concept

• A classifier is a label with a hierarchical format (e.g. programming-language::python) and its associated metadata (e.g. description, maintainer, ...) that is curated by the community
• Classifiers are used to annotate osparc entities (e.g. services, studies, data) in a particular subject area,
• Our approach to classifiers combines two existing implementations: the python distribution metadata standard PEP 301 for the hierarchical-names and a curation workflow very similar to Github Explore.

Examples of Classifiers Labeling

For some inspiration, check:

• python packages are annotated using these classifiers
• Github explore topics

Curation Workflow

The basic idea is to reuse well known versioning, testing,review and integration techniques in software development to continuously curate classifiers

• Every classifier is defined as a directory under data/classifiers that contains files with metadata and resources:
  • the classifier name is given by the folder path
  • index.md is a markdown format file with a Jekyll front-matter block for the classifier metadata
  • other resources (e.g. figures, images) refered in the index.md file
• Any changes or new classifiers are pull requested (PR) to this repository.
• Any PR is subject to the community review and triggers automatic validation of the changes/classifiers
  • Validation is based on data schemas or agreed conventions
  • The PR is only accepted when a minimum number of reviewers accept it and the automatic validation passes
• When PR is accepted, it is merge into the main branch AND the continuous integration transforms the classifiers into a bundled artifact that can be consumed by osparc.io

This workflows guarantees continues versioning, validation and community review of the classifiers in addition to produce a osparc.io-compatible bundle artifact with any version of the classifiers.

Organization of the Repository

• .github : all github and github actions configuration to trigger the automatic validation and artifacts generation.
• .vscode : some template configs for one of the coolest open IDEs in the world ;-)
• data/classifiers: where classifiers names and metadata is hierarchicaly stored
• tools : some python code to test data based on some json-schemas and layout conventions.

References

• A gentle intro to github workflows used here for data curation