The Metadata Landscape: Conventions for Semantics, Syntax, and Structure in the Internet Commons

1
The Metadata LandscapeConventions for
Semantics, Syntax, and Structure in the Internet
Commons
Interfaces to Scientific Data Archives Pasadena,
California

• Stuart Weibel
• Senior Research Scientist
• OCLC Office of Research
• March 24-27, 1998

2
Outline of Todays Talk

• Motivations for developing new conventions for
  resource description
• The Dublin Core Metadata Initiative Semantics
  for resource description
• The Resource Description Framework encoding and
  transport for Web metadata

3
Metadatastructured data about data
A resource description community is characterized
by common semantic, structural, and syntactic
conventions for exchange of resource description
information
4
The Internet Commons embraces many formal and
informal Resource Description Communities
5
Interoperabilityrequires conventions about

• Semantics
• The meaning of the elements
• Structure
• human-readable
• machine-parseable
• Syntax
• grammars to convey semantics and structure

6
The Dublin Core Metadata Workshop Series

• How to improve resource discovery on the Web?
• simple resource description semantics
• Build an interdisciplinary consensus about a core
  element set for resource discovery
• simple and intuitive
• cross-disciplinary
• international
• flexible

7
Metadata Workshop Series and
Related Events

• Chicago WWW Conference Oct,
  1994
• OCLC/NCSA Metadata Workshop Mar, 1995
• OCLC/UKOLN Warwick Workshop April,
  1996
• W3C Dist. Indexing and Searching
  May, 1996
• CNI/OCLC Image Metadata Workshop Sep, 1996
• DC-4, Canberra, Australia Mar, 1997
• Meta Access Summit at RLG
  July, 1997
• EPA Metadata Registry Workshop
  July, 1997
• DC-5, Helsinki, Finland Oct, 1997

8
The Dublin CoreMetadata Element Set

• Title
• Author/Creator
• Subject /Keywords
• Description
• Publisher
• Other Contributor
• Date

• Resource Type
• Format
• Resource Identifier
• Source
• Language
• Relation
• Coverage
• Rights Management

9
Central Characteristics of the
Dublin Core Metadata Element Set

• Descriptive metadata for resource discovery (15
  elements)
• All elements optional
• All elements repeatable
• Extensible (a starting place for richer
  description)
• Interdisciplinary (semantic interoperability)
• International (10 languages and growing)

10
Extensibility

• Refined semantics (Ukrainian Doll model)
• improve sharpness of description with qualifiers
  that refine semantics (controlled vocabularies,
  encoding standards)
• Extended semantics (Lego block model)
• additional elements
• complementary packages of metadata
  (administrative, rights management,
  discipline-specific, etc)

11
What might Extensibility mean for Scientific
Data Communities?

• High-level descriptors to describe data sets
• Use of domain-specific schemes
• Schemes to refine the semantics of Subject,
  Description, Format, Relation, Coverage.
• Controlled vocabularies
• There is no magic If you couldnt play the piano
  before your operation, you wont be able to
  after.

12
Steps Toward Standardization

• IETF informational RFCs of Dublin Core semantics
  and syntax for unqualified DC
• NISO standardization initiated
• ISO standardization is under consideration as
  well

13
DC Implementation Projects

• 50 major implementation projects in 10 countries
  (see the DC home page)
• Australian Government Locator Service
• Danish Online Government Information and Danish
  National Bibliography
• Corporate interest in document management
  semantics
• eg. Boeing, Ford, Nokia
• Metadata for Digital Object Identifiers (DOIs)

14
Why consider the Dublin Core?

• You have a rich standard, need a simple one
  (probably for cost reasons)
• You want to reveal your data to other communities
  (via the Web) using commonly understood semantics
• You want to provide unified access to databases
  with different underlying schemas
• You need core description semantics and dont
  feel compelled to invent them anew

15
Resource Description FormatAn architecture for
metadata on the Web

• W3C Initiative (Formal Working Group)
• Conventions to support interoperability among
  applications that exchange metadata
• Syntax expressed in XML
• Semantics defined by others
• Promote expression of semantics in syntax that
  can be processed by machines as well as humans

16
RDF Working Groups

• Model and Syntax Working Group (baked)
• conventions for encoding arbitrary varieties of
  metadata (semantics defined by others)
• Schemas Working Group (half-baked)
• conventions for defining interoperable schemas
• Search Protocols (looking for the recipe)
• conventions for indexing and searching protocols

17
The RDF Data Model
18
RDF (pseudo) Syntax Example
lt? XMLNamespace HREF http//purl.org/RDF/RDFCo
re AS RDF ?gt lt? XMLNamespace HREF
http//oclc.org/DublinCore AS DC
?gt ltRDFDescription HREF http//purl.oclc.org
/metadata/dublin_coregt ltDC gt
lt/gt ltDC
gt
lt/gt ltDC gt
lt/gt ltDC gt

lt/DC
gt lt/RDFDescriptiongt
Title
Creator Date Subject Subject
19
RDF Why is it important?

• Market demand for deployment
• Software infrastructure will be ubiquitous
• RDF provides a model and syntactical framework
  for metadata in this infrastructure
• Will support independently developed and
  maintained metadata element sets
• e.g. MARC, DC, TEI, EAD, CIMI, GILS, IAFA,
  content ratings...

20
XML-Datacomplementary or competitive?

• W3C Technical Note released in January 98
• Provides for strong data typing within XML
• Semantic and syntactic schemas within XML
  (overlapping RDF functionality)
• Tim Berners-Lee (Feb 10, 1998)
  The relationship between the roles of XML
  level (structural) schemas and RDF level
  (semantic) schemas is not yet clear.

21
Finally...

• We will have means for expressing highly
  structured data and metadata on the Web
• Tools will be integrated into Web infrastructure
  for creating and managing metadata (possibly this
  year)
• The foundations for extensible semantics are in
  place
• The biggest challenge is to promote consistent
  deployment

22
Additional Information on Dublin Core and RDF

• Dublin Core Homepage
• http//purl.org/metadata/dublin_core
• RDF Working Group Home Page
• http//www.w3.org/RDF