A slightly opinionated, common set of 'rules' for creating rdf data using 'rdf-engine':

• Data Rules: for loading tables (csv), hierarchical (json), and rdf (ttl).
• Mapping rule: SPARQL construct
• Ontology rules: Inference and validation using TopQuadrant

These rules come together in the 'engine'.

Development

Develop with uv sync --all-packages --all-extras.

Run python tasks.py stamp_ver before pypi publishing.

Design Choices

These are choices given the common use case of mapping data to an ontology. They are somewhat firm.

• RDF1.2 annotates tripes with metadata as <<?s ?p ?o>> ?mp ?mo where ?mp and ?mo correspond to simple (key,value) pairs of metadata.
• Mappings are in the form of SPARQL constructs stored as files with a .mapping.rq extension (can also be .mapping.sparql).
• Each (specified) ontology will be processed separately.

Usage

Examine how rules are fed into an engine in the tests. The common workflow is:

1. Specify rules. Each rule is constructed by providing arguments to mkrule. mkrule processes objects such as file paths and data objects. A distinction is made for 'data rules'; These are only triggered once in the beginning.

2. Specify engine run parameters. See run function documentation.

3. Extract data subsets with queries. 'System' queries are mapped_and_inferred and validation (results). It's best to extract these using the query function as they slightly depend on configuration.