Metadata for images Michael Day UKOLN: UK Office for Library and Information Networking University of Bath m.day@ukoln.ac.uk.

Metadata for images

Michael Day

UKOLN: UK Office for Library and

Information Networking

University of Bath

http://www.ukoln.ac.uk/

m.day@ukoln.ac.uk

Metadata for images

Michael Day

The Challenge of Image Retrieval

CIR 99 - Second UK Conference on Image Retrieval,

Forte Posthouse Hotel, Newcastle upon Tyne, 25-26 February 1999.

http://www.unn.ac.uk/iidr/conference.html

3

Presentation Outline

Metadata:• Contexts• Dublin Core initiative• Resource Description Framework

Distributed and heterogeneous information:• CIMI• European initiatives

Other metadata applications:• Representation and authentication• Digital preservation

4

Metadata

Contexts:• Rapidly growing corpus of image-

based information• CBIR and metadata• Metadata = data about data or “… the

Internet-age term for structured data about data” - Joint NSF-EU Working Group on Metadata (1998)

• Format diversity

5

A metadata typology

Simple Rich

Adapted from: Lorcan Dempsey and Rachel Heery, “Metadata: a current view of

practice and issues”, Journal of Documentation, vol. 54, no.2, March 1998,

pp. 145-172.

Band One Band Two Band Three

(full textindexes)

(simplestructuredgenericformats)

(more complexstructure,domainspecific)

(part of largersemanticframework)

Proprietaryformats

ProprietaryformatsDublin CoreROADSIAFA/Whois++templates

FGDCMARC

TEI headersICPSREADCIMI

6

Types of metadata

Format diversity likely to persist

Metadata creation and cataloguing

Subject classification schemes:• ICONCLASS

Thesauri• Art and Architecture Thesaurus (AAT)

7

Dublin Core (1)

International initiative to define a core set of metadata elements for resource discovery on the Internet

• Six DC workshops (to date):• DC-1 (Dublin, Ohio) - 1995• DC-2 (Warwick) - 1996• DC-3 (Dublin, Ohio) - 1996• DC-4 (Canberra) - 1997• DC-5 (Helsinki) - 1997• DC-6 (Washington, D.C.) - 1998• DC-7 (Frankfurt am Main) - 1999

http://purl.oclc.org/dc

8

Dublin Core (2)15 Elements:

• Title • Subject • Description • Creator • Publisher • Contributor • Date • Type

Core elements defined in RFC 2413:

http://src.doc.ic.ac.uk/computing/internet/rfc/rfc2413.txt

• Format • Identifier • Source • Language • Relation• Coverage • Rights

9

Dublin Core Qualifiers

TYPE - refines the meaning of elements:– Relation TYPE=IsPartOf

SCHEME - associates the value with an externally defined ‘scheme’:

– Subject SCHEME=DDC– Date SCHEME=ISO 8601

LANGUAGE - indicates the language of the value:

– Title LANGUAGE=en

10

Dublin Core syntax

Syntax issues:• Simple DC can be embedded into

HTML Web pages– Limited functionality

• Web moving to Extensible Markup Language (XML)

• Resource Description Framework– RDF ... “an architecture for metadata on

the Web”

11

RDF

Resource Description Framework:• World Wide Web Consortium (W3C)• Data model and XML based syntax• An implementation of the conceptual

‘Warwick Framework’• Modular interoperability• Useful for aggregating the different

metadata types required for managing digital information over time

http://www.w3.org/RDF/

12

Integrating access

Distributed and heterogeneous information:• ANSI/NISO Z39.50 protocol

Applications:• Computer Interchange of Museum

Information (CIMI) Consortium• Aquarelle project• Electronic Library Image Service for

Europe (ELISE)• Arts and Humanities Data Service

(AHDS)

13

Research processes

Metadata interacts with the research process: • Discovery• Retrieval• Collation• Analysis• Re-presentation

David Bearman and Jennifer Trant, Unifying our cultural memory. Information Landscapes for a Learning Society: Networking and the Future of Libraries 3, University of Bath, 29 June - 1 July 1998.

http://www.archimuse.com/papers/ukoln98paper/index.html

14

Standards for images

Types (Howard Besser):• Technical information (for viewing)• Capture processes• Quality and veracity• Original object• Authentication• Rights metadata

Where should this metadata be kept?• Image headers• Separate databases

15

Making of America II

Three types of metadata:• Descriptive• Structural• Administrative

16

Digital preservation

The existence of relevant metadata is the key to the future utilisation of image-based information

Preservation strategies depend upon metadata:

• Digital Rosetta Stone (DRS)• "Super-metadata"

17

Research Libraries Group

• Date• Transcriber• Producer• Capture device• Capture details• Change history• Validation key• Encryption• Watermark

• Resolution• Compression• Source• Color• Color management• Color bar / Grey

scale bar• Control targets

RLG Working Group on Preservation Issues of Metadata (1998)

18

OAIS

A high-level model for ‘archival information object classes’:

• Content Information• Preservation Description Information

– Reference Information

– Context Information

– Provenance Information

– Fixity Information

• Packaging Information• Descriptive Information

19

Implementations

National Library of Australia• PANDORA project• 'logical data model'

Cedars project• Electronic Libraries Programme• Consortium of University Research

Libraries• Defining data elements• Demonstrators

20

Conclusions

Metadata complements CBIR approach to image retrieval

Metadata has wider applications than discovery and retrieval

• Representation of information• Rights management• Authentication• Preservation

21

UKOLN

UKOLN is funded by the British Library Research and Innovation Centre (BLRIC), the Joint Information Systems Committee (JISC) of the UK Higher Education Funding Councils, as well as by project funding from the JISC’s Electronic Libraries (eLib) Programme and the European Union. UKOLN also receives support from the University of Bath, where it is based.

http://www.ukoln.ac.uk/