sparql-to-graphviz

When you find unknown data in a SPARQL endpoint, do you also go:

SELECT DISTINCT ?class
WHERE {
  [] a ?class .
}

Or:

SELECT DISTINCT ?property
WHERE {
  [] ?property [] .
}

Yeah, me too.

This tools helps you to explore an unfamiliar dataset in a SPARQL endpoint, optionally restricted to a named graph. It produces a class diagram in the DOT language, which can be turned into images by Graphviz. The class diagram shows an empirical schema of the explored dataset. The schema reflects the structure of instance data in terms of its vocabularies.

If you want a more feature-rich and interactive visualization, give LODSight a try.

Warning: The tool can also hurt your feelings by showing what a mess your data is.

Usage

Compile using Leiningen:

git clone https://github.com/jindrichmynarz/sparql-to-graphviz.git
cd sparql-to-graphviz
lein bin

Observe the offered parameters:

target/sparql_to_graphviz --help

Find out what is hiding in your local RDF store:

target/sparql_to_graphviz -e http://localhost:8890/sparql |
  neato -Tsvg -Gepsilon=.001 -o localhost.svg

Give it some time, while it runs a boatload of SPARQL queries. When it finishes, localhost.svg should contain something beautiful. Exporting to SVG also allows you to make the visualization fancier using vector graphics tools.

For example, the resulting mess can look like this:

Caveats

• If there are no instantiated classes, the program fails.
• Classes identified with blank nodes are ignored.

License

Copyright © 2017 Jindřich Mynarz

Distributed under the Eclipse Public License version 1.0.