EKGF tool
Triplify
Turn CSV files and spreadsheets into RDF, preserving their original structure and values.
Capture your source data without defining a mapping first. Use SPARQL and SHACL afterward to transform, connect and validate it.
What it does
triplify ./dumps -o raw.nq.gz
Triplify walks the files and directories you give it, honouring
.gitignore, and captures every table it can read as RDF. It reads CSV, TSV,
XLSX, XLSM, XLSB, XLS, ODS and Parquet files. The output follows the
EKGF Raw Data Ontology:
sources, views, columns, rows and values, exactly as they physically are.
Each source gets its own named graph, along with its metadata: a DCAT distribution, an SPDX SHA-256 checksum, its Git origin, and a PROV activity for the run. See Output model for the details.
Principles
Capture, never map
Readers report physical structure and the source's own value types. Triplify never infers datatypes from text, never picks key columns and never renames anything. Mapping happens later, in the graph, with SPARQL and SHACL.
Content-addressed identity
Every IRI is a urn:uuid: that anyone can recompute from the source bytes and
a position. Identical files found twice are triplified once.
Provenance included
Every source records its media type, byte size, location, checksum and Git
blob. Inside a Git work tree it also records the repository, commit and path,
and whether the file differs from HEAD.
Built for big dumps
Files are processed concurrently, one per core, into a single writer. Output streams, optionally gzip-compressed, unless you ask for canonical RDF.
Learn more
- Why capture first explains the approach and how it differs from mapping languages.
- Usage is the command-line reference.
- Contributing covers the architecture, conventions and releases.