Download our brochure: Mapping Workbench

Stop rewriting your mappings every time a source schema changes. Mapping Workbench is an IDE where semantic engineers manage mapping rules, built on our published method: rules agreed once, then versioned and tested across every schema version.

This one-page brochure shows the method, the numbers behind it and where it fits in your stack.

What is inside

Proven across 18 schema versions. We built Mapping Workbench to map the Publications Office’s procurement notices, legacy Standard Forms and current eForms, into the eProcurement Ontology: more than 1,500 source fields, more than 40 document types, 18 schema versions (5 Standard Forms, 13 eForms) and ontology versions 3 to 5.

Our method, in four steps:

• Conceptual mapping: domain experts and semantic engineers agree, field by field, what each source element means in the target model.
• Technical mapping: engineers implement each agreed rule as an executable mapping, in RML, YARRRML or code.
• Validation: SHACL and SPARQL tests are generated from the conceptual mapping, and coverage reports show which source fields are mapped.
• Packaging: each version ships as a self-contained mapping package, with generated documentation.

Where it fits: Mapping Workbench manages, tests and versions the rules; each release is one versioned, documented mapping package; your orchestrator (Airflow, n8n or your own) runs and deploys it; your shared model receives validated data, ready for your knowledge graph. We can set up and operate the pipeline for you.

Source formats: XML sources work today; JSON, CSV and SQL come next.

Open source: AGPL-3.0, with a commercial licence available. Presented at SEMANTiCS 2024.

Who it is for

Semantic engineers and data teams with many heterogeneous sources and one target model. In life sciences, the same method applies to your lab notebook, LIMS and assay exports.

Talk it through with us

The brochure is the summary; the useful part is the conversation that follows it. Bring a sample XML export (synthetic data is fine), and we map a few fields with you, live.