Dublin Core metadata in CMS frameworks

About the Presentation

The perspective of embracing the DC metadata standard for a large project is daunting, in particular the first time. And the second. And the third. By the fourth we are starting to feel we know what we are doing. Unfortunately, most Open Source CMSs do not provide yet a solid metadata management framework.

We aim to explore the specific needs of a metadata framework in the context of a CMS. Where does it fit in the (content) object model, in the workflow, in the version control framework, in the UI? How to harness the extra functionality that metadata can offer?

These questions came to light in a range of ground-breaking projects we have developed over the last 4 years. The answers vary with each project -- and mature CMSs need a metadata framework that can provide those answers.

We are currently adding metadata extensions to the Midgard CMS framework, that make it easy to implement DublinCore-based metadata application profiles. We will also discuss a standalone DublinCore metadata framework that provides repository, search engine and editor.

The presentation items will be illustrated with examples from existing projects. There will also be a brief tour of the Midgard implementation internals and the architectural choices, as well as the standalone metadata framework.

Presentation outline

• Intro
• Who we are
• Involvement in DC and CMS related projects
• Data vs Metadata
• 'Content objects'
• Integrated/separate storage and management of content and metadata
• Workflow
• Edit metadata as a separate stage?
• Separate versioning?
• Metadata-only records and external metadata feeds and
• Granularity
• Collections and resources
• Resource parts and versions
• Associated resources
• Metadata values
• Value Qualifiers and "freetext"
• Enforcing VQs -- formats, CVLs and thesauri
• The advantages of using thesauri
• Searching metadata
• Engines: custom vs off the shelf
• TKI search engine
• Lucene
• UI: complex vs complex
• Matching user's knowledge and expectations
• Using thesauri and VQs to search
• Multiple languages
• Multilingual metadata
• Workflow
• Search UI
• Cross-linking thesauri
• Storage
• Text files (XML/RDF) with RCS/CVS control
• RDBMS
• Metadata editors
• VQs and validation
• Hosting 3rd party metadata and its impact on the ob
• Publication
• HTTP Headers
• HTML Headers
• XML/RDF
• Where are the spiders?

We are aware of that the presentation as outlined exceeds the 45 minutes available, and we expect to trim or abridge the less interesting topics. Feedback is welcome in this regard.

About the projects

Te Kete Ipurangi - A huge bilingual educational portal and web community. Includes a complex DublinCore metadata repository and search system using LAMP. Includes an advances metadata search UI <http://www.tki.org.nz/e/search/>. Uses CVS for Content Management.

NZ eGovernment Portal - A portal to an large backend metadata repository based on NZGLS, a DC extension. Uses Midgard as Content Management framework. <http://www.govt.nz/>

Many other projects involving DublinCore and CMSs: NZ Encyclopedia, Unesco and SOCCI.

About the speakers

Martin Langhoff - Technical architect and lead programmer for CWA New Media's flagship metadata projects. Also involved in projects within the open source community, including Midgard and several Perl modules. <http://www.cwa.co.nz/people/martin.html>

David Copeland - Director at CWA New Media and metadata advocate. David leads CWA's involvement in the SOCCI project, which is creating learning objects and learning management systems that come together using DC-based metadata. <http://www.cwa.co.nz/people/david.html>

Trackback