LOD Cloud VoID Generator

This project is about generating a VoID description of all the datasets in the LOD Cloud diagram. It currently generates an RDF dump containing these descriptions, available online here:

http://lod-cloud.net/data/void.ttl

Background

The LOD Cloud diagram is a pictorial of datasets published in linked data format on the Web.

Metadata for these datasets is recorded in the Data Hub, an open directory of datasets. The lodcloud group contains metadata for all the datasets in the LOD Cloud diagram.

VoID is an RDF vocabulary for expressing metadata about such datasets in RDF format.

Running it

Clone the repository and fetch required dependencies:

git clone https://github.com/lod-cloud/datahub2void.git
cd datahub2void
git submodule update --init

Run the code:

php generate.php

This takes a few minutes. It creates a file void.ttl in the current directory.

Details

This uses Ckan_Client-PHP for accessing the CKAN API.

This uses some code taken from Neologism and DBpedia to serialize Turtle via ARC2. This code is found in rdfwriter.inc.php. The class offers an API that's a bit nicer than ARC2's triple representation, it fixes some bugs related to literal serialization in ARC2's TurtleSerializer, and tweaks the layout of the produced Turtle to provide (subjectively) nicer-looking Turtle output.

To Do

• Automatically publish the VoID file to lod-cloud.net
• Fetch all the data with a single API call instead of one per dataset
• Add a VoID description for this dataset itself
• Better validation for URIs, triple numbers, etc
• Better/other vocabulary for tags?
• Interpret some more of the tags?
• Do something with sparql_named_graph custom field
• Do something with other custom fields
• Do something with version field
• Do something with the ratings
• Better consolidation of authors/maintainers
• consolidate the fixed TurtleWriter and contribute back to ARC

History

At some point, this project was intended to produce not just an RDF dump, but also RDF and HTML descriptions of each entity described in the dump. That effort stalled, and is removed from the current codebase, but can still be found in the feature-html branch.

Acknowledgements

Originally created by Richard Cyganiak (richard@cyganiak.de).

Thanks to Michael Hausenblas for feedback and comments.

Thanks to the LOD community for publishing all these datasets, and thanks to OKFN for hosting the metadata!