Son of Dublin Core (SoDC) — Encoding, Validating and Harvesting Graph-based (Meta)data — Version 0.2 (alpha)

Being at the Dublin Core conference last week gave me some ideas, especially about how to bring the DCMI architecture, RDF, OAI-PMH and OAI-ORE closer together. I gave those ideas a name — “Son of Dublin Core (SoDC)” — and I hacked up a proof of concept.

I’ve just done a few fixes and extensions, and released SoDC version 0.2 (alpha). The latest release will always be available from http://purl.org/net/sodc/. SoDC is just an idea at this stage, so let me know what you think…

There’s two main parts to SoDC…

“Vocabularies” Tutorial for DC-2007 — Preview

Here’s a preview of the tutorial on “vocabularies” I’m giving next week at the International Conference on Dublin Core and Metadata Applications:

• http://isegserv.itd.rl.ac.uk/public/ajm65/dc2007/tutorial.pdf

The link will always give you the latest version – I may do some minor edits before next week.

The scope of the tutorial is restricted to controlled structured vocabularies for retrieval. I’ve tried to cover some basic topics, such as a simple common model for different vocabulary types, the business context for vocabularies, and the use and re-use of vocabularies in a networked environment (i.e. on the Web).

I’ve also tried to go beyond previous tutorials at the Dublin Core conference, and address the use of controlled structured vocabularies in retrieval systems, including the use of vocabulary mappings to implement retrieval across heterogeneous metadata.

There are also files in open office and powerpoint files, which can all be downloaded from:

• http://isegserv.itd.rl.ac.uk/cvs-public/ajm65/dc2007/

Feel free to re-use slides or diagrams with attribution.

RDA: Resource Description and Access

This week I attended a meeting at the British Library, concerning a replacement for the Anglo-American Cataloguing Rules. The replacement under development is called “RDA: Resource Description and Access”. The meeting was set up to explore the links between RDA, DCMI‘s metadata standards and the Semantic Web technology stack, in order to better meet requirements for discovery and access to library resources in a networked environment. More details of the meeting can be found at the meeting wiki page.

The meeting was between members of the Dublin Core Metadata Initiative (Diane Hillman, Tom Baker, Andy Powell, Mikael Nilsson), representatives of the JSC (Barbara Tillet, Gordon Dunsire, Robina Clayphan), and the editor of the RDA (Tom Delsey). I was asked to attend because of my work with Semantic Web, metadata and library communities, and gave a short presentation on SKOS.

The main outcome of the meeting was a proposal to jointly develop a new Dublin Core Application Profile for libraries, based on the RDA and on FRBR. The profile would also be closely linked to the ePrints application profile developed in the UK for Intute. The ePrints AP is an example of how the Dublin Core community is moving beyond the “15 elements” towards providing a general framework to support rich and highly structured metadata, via the Dublin Core Abstract Model [latest draft].

DCMI Abstract Model and RDF

I read the new draft of the DCMI Abstract Model – this is a “DCMI proposed recommendation” and the comment period expires in the next couple of days. I wrote a review of the document, in which I said that DCMI should specify an abstract syntax for DCMI metadata, and a mapping from that abstract syntax to RDF graphs, rather than develop an “abstract model”.

It occurred to me that the purpose of DCMI should be to deal with three fundamental concerns of metadata consumers. Firstly, processing metadata in different concrete syntaxes (syntactic interoperability). Secondly, merging metadata from different sources (low-level semantic interoperability). Thirdly, making valid inferences (higher-level semantic interoperability). I argued that DCMI should specify an abstract syntax (this would deal with the syntactic interoperability) and should specify a mapping from the abstract syntax to RDF graphs (this would deal with semantic interoperability, because both merging and valid inferences are already provided by the RDF semantics).

To put my foot where my mouth is, I wrote an abstract syntax for DCMI metadata and a mapping to RDF graphs.