A standards-based POC kit for turning relational databases into ontology-backed RDF knowledge graphs.
Make the boring path boring.
Getting started: see SETUP.md for a step-by-step Windows install and an end-to-end trial run against the bundled Chinook database.
rdb2kg is intended to make the standard relational-to-RDF path easy to demonstrate, test, and explain.
Given a relational database and an R2RML mapping, it should be easy to:
- inspect the database schema,
- generate or edit a mapping,
- materialize RDF triples,
- load them into a local RDF store,
- run SPARQL queries,
- compare expected results,
- and explain what happened.
The goal is not to invent a new graph model, a new query language, or a new mapping vocabulary.
The goal is to make the existing standards usable for ordinary proof-of-concept work.
A company should be able to try a knowledge graph over an existing relational database without replacing the database, buying a platform, or hand-writing a one-off ETL pipeline.
The standards already exist:
- relational databases store the data,
- R2RML describes how relational data maps to RDF,
- RDF provides the graph data model,
- ontologies describe the meaning of the data,
- SPARQL queries the graph.
This project exists because that path should be easier than it usually is.
The intended workflow is:
relational database
+ ontology
+ R2RML mapping
↓
materialized RDF triples
↓
local RDF store
↓
SPARQL queries
↓
explainable POC