Note: This report was originally published in D-Lib Magazine, June 1997, by Stuart Weibel (OCLC), Renato Iannella (DSTC), and Warwick Cathro (National Library of Australia).
Original publication: https://www.dlib.org/dlib/june97/metadata/06weibel.html

The 4th Dublin Core Metadata Workshop Report

Stuart Weibel, OCLC; Renato Iannella, DSTC; Warwick Cathro, National Library of Australia

Introduction

The fourth Dublin Core Metadata Workshop convened March 3–5, 1997, at the National Library of Australia in Canberra. Sixty-five participants from 12 countries across 4 continents gathered, representing digital library researchers, Internet specialists, content specialists, and librarians.

The workshop was co-sponsored by the National Library of Australia, the Distributed Systems Technology Centre (DSTC), and OCLC Online Computer Library Center.

Central Issues

The workshop addressed three primary concerns:

1. Element structure — possible qualifiers for syntax specification
2. Extensibility mechanisms — for core metadata sets
3. Element refinement — particularly for the coverage, relation, and rights management elements

Minimalists vs. Structuralists

The workshop exposed philosophical differences between two pragmatic camps:

Minimalists prioritized simplicity to prevent semantic drift. They argued that a simpler standard would see wider adoption and that complexity could be managed through external mechanisms.

Structuralists accepted flexibility risks for enhanced community-specific functionality. They argued that real-world deployment across different domains required richer qualification mechanisms.

This tension was not a conflict to be resolved but a productive dialectic. The Dublin Core's subsequent evolution reflected a careful balance between these positions.

The Canberra Qualifiers

Four qualifier types were formalized:

Language. Specifies the language of the descriptor itself (distinct from the resource Language element). This supports multilingual metadata creation.

Scheme. Provides interpretation context for element values. For example, a Subject element might use LCSH (Library of Congress Subject Headings) or DDC (Dewey Decimal Classification) as its scheme.

Type. Narrows field semantics as sub-element names. For example, the Date element might be qualified with a type of "created" or "modified."

Authority Attributes. Link qualifiers to controlled namespaces, enabling machine-readable identification of qualification schemes.

HTML Implementation Approaches

Two competing methods for encoding Dublin Core in HTML emerged:

Overloaded Content embedded qualifiers within the META tag CONTENT attribute. This approach worked with existing HTML but mixed data with metadata about the data in a single field.

Additional Attribute used unofficial SCHEME and LANG attributes in the META tag for cleaner representation. This was more elegant but required HTML extensions not yet standardized.

The debate between these approaches reflected the broader minimalist-structuralist tension: simplicity of adoption versus richness of expression.

Element Set Evolution

The Dublin Core evolved from the original 13 elements (DC-1) through 15 elements (DC-3) to a stabilized 15-element set with modified and more precise names. This marked a transition from a development phase to a deployment phase — the core set was now considered stable enough for implementation.

Working Groups

Coverage Element. A position paper on geographic and spatial data applications was developed, addressing the needs of the GIS community while maintaining compatibility with the Dublin Core's simplicity goals.

Multilinguality. Addressed internationalization challenges, particularly the distinction between the language of the metadata itself and the language of the described resource.

Metadata Registries. Tackled namespace management and sub-element authority — the question of who controls the definition and evolution of qualifiers and extensions.

Planned Deliverables

Four RFC (Request for Comments) documents were planned for IETF standardization:

1. Dublin Core semantics for simple resource description
2. HTML encoding conventions for unqualified Dublin Core
3. Qualified metadata semantics
4. HTML encoding for qualified metadata

Web Architecture Context

The workshop discussions took place against a backdrop of rapidly evolving web technologies:

• XML (Extensible Markup Language) — promising a more structured alternative to HTML for metadata encoding
• WebDAV — extending HTTP for resource metadata and version control
• Web Collections — an XML application for describing object properties
• PICS-NG — a label infrastructure for resource description
• Cougar — HTML enhancement work

Steering Committee

The Dublin Core Steering Committee included: Tom Baker, Lorcan Dempsey, Ricky Erway, Juha Hakala, John Kunze, Carl Lagoze, Clifford Lynch, Andreas Paepke, Frank Roos, Diann Rusch-Feja, Andrew Wells, and Bernal Rajapatirana.

Next Steps

DC-5 was tentatively scheduled for October 6–8, 1997, in Helsinki, Finland, focusing on convergence among early implementers and syntax standardization.