Moving from a locally-developed data model to a standard conceptual model Jenn Riley Metadata...

Post on 13-Jan-2016

218 views 0 download

Tags:

Transcript of Moving from a locally-developed data model to a standard conceptual model Jenn Riley Metadata...

Moving from a locally-developed data model to a standard conceptual modelJenn RileyMetadata LibrarianIndiana University Digital Library Program

I’m a practitioner.

•And a librarian.•But I work in a department whose mission

is to advance the state of the art in digital libraries,

•and I’m particularly interested in innovative discovery systems.

Therefore, I often act as a bridge between the researcher and the implementer.

August 6, 2008

International Society for Knowledge

Organization

2

Libraries’ metadata focus• Practical!• Element sets and the records that implement

them• Metadata element sets tend to be defined by

their encodings▫Rarely do element sets from this community

have multiple encodings▫Rarely is there an externally defined model on

which the encoding is based• Only recently has this community started

thinking about conceptual models

August 6, 2008

International Society for Knowledge

Organization

3

Some conceptual models•FRBR: Functional Requirements for

Bibliographic Records, 1998 report from the International Federation of Library Associations and Institutions (IFLA)

•CIDOC CRM: International Committee for Museum Documentation Conceptual Reference Model, ISO 21127:2006

•DCMI Abstract Model, 2007▫“Information model”▫At a higher level of abstraction than the

first two

August 6, 2008

International Society for Knowledge

Organization

4

What’s the connection?

•Significant literature on both topics, but they rarely reference one another

•Should also note that the categories “element set” and “conceptual model” don’t have strict boundaries

•Does a metadata element set need to be explicitly based on a conceptual model?

•What does it even mean for an element set to conform to a conceptual model?

August 6, 2008

International Society for Knowledge

Organization

5

What the community has realized•An element set necessarily instantiates an

underlying conceptual model▫Even if it’s not explicitly defined▫Even if it’s internally inconsistent, or not

really what was intended•The conceptual model has a profound

effect on what can be done with the metadata, and what can be described with it

•Mapping between element sets easier when they use the same conceptual model

August 6, 2008

International Society for Knowledge

Organization

6

DCMI leading work in this area• DCMES, 1995• Warwick Framework (format-independent

container architecture) and slight revisions, 1996

• Introduction of qualifiers in 2000• DCMI Abstract Model

▫First draft 2004▫Current version June 2007

• Encodings▫Have changed over time▫DCMI has long presented several options▫Now will be explicitly connected to the Abstract

Model• Libraries should learn from this development

August 6, 2008

International Society for Knowledge

Organization

7

Applying these principles to Variations @ Indiana University• Variations Digital Music Library in operation since

1995▫ Streaming audio▫ Scanned scores, and a few encoded scores

• Work-based data model developed in 2001▫ Defined as a human-readable data dictionary▫ Data structured as XML inside the system▫ XML Schema for Java classes to interact with not

developed until 2005• Current work focused on long-term sustainability

▫ Locally-developed data model is a liability▫ FRBR gaining real traction in the library community▫ It became obvious we needed a change, and one based

on conformance to a standard conceptual model

August 6, 2008

International Society for Knowledge

Organization

8

Original Variations data model

August 6, 2008

International Society for Knowledge

Organization

9

Variations vs. FRBRVariations 2/3 Entity FRBR Group 1

EntityWork (more concrete than FRBR Work)

Work

Instantiation (can only appear on one Container)

Expression

Container (includes some copy-specific data)

Manifestation

Media Object (defined as a digital file)

Item

August 6, 2008

International Society for Knowledge

Organization

10

Current status of our work• Reports analyzing FRBR/FRAD as applied to

music▫Music-specific entity definitions▫Attributes needed/not needed▫Relationships needed/not needed▫Additions to FRBR/FRAD needed

• Currently investigating encodings▫No data structure from IFLA, and other library

bodies haven’t stepped up▫ Internal data representation vs. export formats

• Pending grant application for development work to perform the switch – stay tuned!

August 6, 2008

International Society for Knowledge

Organization

11

Encodings under consideration (1)• FRBR in RDF

▫Researcher-driven▫No stable body behind it▫Only covers entities and relationships, not

attributes• FRBRoo

▫“Harmonization” of FRBR and CIDOC/CRM▫Limits Events to those for Group 1 entities▫“Electronic publishing” model doesn’t include a

Manifestation▫No OWL ontology for FRBRoo yet, only for

CIDOC/CRM

August 6, 2008

International Society for Knowledge

Organization

12

Encodings under consideration (2)•Music Ontology

▫Scope considerably wider than what Variations needs

▫Lacks model for FRBR Group 3 entities•DCMI/RDA Vocabularies

▫Because RDA is FRBR-based▫But likely not close enough to FRBR for us

•So we may have to make our own▫But would still export some of these other

alternatives

August 6, 2008

International Society for Knowledge

Organization

13

Thank you!

•Let’s find more ways for researchers and practitioners to work together.

•Questions?•For more information:

▫jenlrile@indiana.edu▫These presentation slides:

http://www.dlib.indiana.edu/~jenlrile/presentations/isko2008/isko2008.ppt

August 6, 2008

International Society for Knowledge

Organization

14