FAIR Metadata generator

This Python library can be used from the commandline, it automatically generates descriptive metadata about a RDF knowledge graph (classes, instances, relations between classes).

We run SPARQL queries against the SPARQL endpoint provided by the user to produce HCLS dataset statistics for the dataset in the RDF Turtle format.

Prerequisites

• Python 3.6 or higher, with pip
• Docker (optional)

Installation

Provide instructions to install the package

Install directly from GitHub to try it:

pip3 install git+https://github.com/MaastrichtU-IDS/fair-metadata.git

Or install from source code for development. Using -e means that changes to the source code will be automatically update the package locally.

pip3 install -e .

Usage

Provide working examples on how to run the package

Check the commands available:

fair-metadata
fair-metadata analyze --help

Analyze a SPARQL endpoint

Generate descriptive metadata, about types and relations, for a SPARQL endpoint

fair-metadata analyze https://graphdb.dumontierlab.com/repositories/test-vincent -o metadata.ttl

Create dataset metadata description

Create complete metadata description for your dataset, you will be asked a few questions (such as homepage, license and reference for this dataset)

fair-metadata create -o dataset_metadata.ttl

Run as a Python library

You can also import and use this library in Python:

from fair_metadata.generate_metadata import generate_hcls_from_sparql

sparql_endpoint = 'https://graphdb.dumontierlab.com/repositories/test-vincent'
dataset_uri = 'https://w3id.org/d2s/distribution/default'

g = generate_hcls_from_sparql(sparql_endpoint, dataset_uri)

Continuous Integration

This repository uses GitHub Actions to:

• Automatically run tests at each push to the main branch
  • It uploads the test coverage to SonarCloud (requires to set the SONAR_TOKEN secret)
• Publish the package to PyPI when a release is created (N.B.: the version of the package needs to be increased in setup.py before).

You will need to provide your login credentials using secrets in the repository settings to publish to PyPI: PYPI_USERNAME and PYPI_PASSWORD

Test locally

Install PyTest:

pip3 install -U pytest

Run the tests:

pytest

Run a specific test in a file, and display print in the output:

pytest tests/test_fair_metadata.py::test_create_dataset_metadata -s

Docker

Build the image:

docker build -t fair-metadata .

Run a container:

docker run -it --rm -v $(pwd):/root fair-metadata create -o dataset_metadata.ttl