Title: The Dublin Core Abstract Model
1The Dublin Core Abstract Model a packaging
standard?
- Content Packaging for Complex Objects Technical
Workshop
flickr photo by amalthya
2Why Dublin Core?
this workshop is about content packaging not
metadata!?
DC doesnt do content packaging does it?
DC is just 15 elements for describing Web pages
isnt it?
3DC and content packaging
http//dublincore.org/documents/abstract-model/
- this talk is about the DCMI Abstract Model
- and its relationship to content packaging
- it is not intended as a tutorial
- but I appreciate that the DCMI Abstract Model is
new to many of you - I will therefore start by summarising the
background, context and main features of the DCAM - then Ill give some examples
- and finally try to draw some conclusions
4DCMI Abstract Model background
- in the early days of Dublin Core there was no
explicit model associated with DC metadata
descriptions - there were implicit models and conventional
wisdom - largely flat in nature i.e. a set of metadata
elements describing a single thing (e.g. a Web
page) - and there were known problems
- like sometimes it was obvious that an element was
really being used to describe a second thing
(e.g. the author of a Web page)
5DCMI Abstract Model background
- as the various DC syntaxes matured
- XHTML, XML and RDF/XML
- the underlying model became more important
- primarily as a mechanism for mapping between
syntaxes - and there have been a number of attempts at
applying the RDF model to DC
6DCMI Abstract Model key features
- the DCAM (first published in 2005) attempts to
make explicit the model that underpins DC - the DCAM starts from the central notion of a
description set - a set of descriptions about a group of related
things (resources) - where each description is about a single
resource - and where each description is essentially made
up of property/value pair statements - descriptions sets are instantiated as records
(e.g. using XHTML, XML or RDF/XML) for the
purpose of exchanging information between
networked systems
7Model summary
8DCAM and relationships
- the DCAM is very open about the nature of the
relationships between the resources described in
a description set - whole / part (e.g. book / chapter / section /
page) - physical / digital (painting / digitised
painting) - object / human (document / author)
- conceptual / physical (work / item)
- or all of the above!
- the relationships between things is articulated
in an application model and captured using the
properties specified in an application profile
9Example 1 Book application model
- here is a very simple application model
0..8
hasPart
Book
Chapter
10Example 1 pseudo-XML description set
- ltdescriptionSetgt
- ltdescription resourceURIhttp//example.org/mybo
okgt - ltstatement propertyURIdctermshasPart
valueURIhttp//example.org/chapter1 /gt - ltstatement propertyURIdctermshasPart
valueURIhttp//example.org/chapter2 /gt - lt/descriptiongt
- ltdescription resourceURIhttp//example.org/chap
ter1gt - ltstatement propertyURIdctitlegt
- ltvalueStringgtChapter 1lt/valueStringgt
- lt/statementgt
- lt/descriptiongt
- ltdescription resourceURIhttp//example.org/chap
ter2gt - ltstatement propertyURIdctitlegt
- ltvalueStringgtChapter 2lt/valueStringgt
- lt/statementgt
- lt/descriptiongt
- lt/descriptionSetgt
11Note 1 objects packaged by reference
- note that objects within the package (the
resources described within the description set)
are passed by reference - i.e. their URL is provided
- this is in common with other packaging standards
- passing by value (i.e. embedding the object
in-line) is theoretically possible using the DCAM
rich representation mechanism (but this is not
discussed further here)
12Note 2 - ordering
- the DCAM has no built-in support for ordering
- the model is graph-based rather than being an
ordered tree - for applications requiring ordering, e.g. the
chapters in a book, it would therefore be
necessary to invent properties (e.g.
mysequenceNumber) to capture the ordering as
part of the description
13Eprints application model
http//www.ariadne.ac.uk/issue50/allinson-et-al/
- here is a more complex application model
ScholarlyWork
14Example 2 psuedo-XML
- ltdescriptionSetgt
- ltdescription resourceURIhttp//eprints.gla.ac.u
k/503/gt - ltstatement propertyURIdctitlegt
ltvalueStringgtAttempts to detect
retrotransposition and de novo deletion of Alus
and other dispersed repeats at specific loci in
the human genome lt/valueStringgt lt/statementgt - ltstatement propertyURIeprintisExpressedAs
valueRefexpression1 /gt - lt/descriptiongt
- ltdescription resourceIdexpression1 gt
- ltstatement propertyURIeprintisManifestedAs
valueRefpdfmanifestation /gt - lt/descriptiongt
- ltdescription resourceIdpdfmanifestation gt
- ltstatement propertyURIeprintisAvailableAs
- valueURIhttp//eprints.gla.ac.uk/503/01/Eu_
J._Hum_Gen.9(2)143_.pdf /gt - ltstatement propertyURIeprintisAvailableAs
- valueURIhttp//www.nature.com/ejhg/journal/
v9/n2/pdf/5200590a.pdf /gt - ltdescriptiongt
- lt! descriptions of the two copies here --gt
- lt/descriptionSetgt
15Note 3 - Compound vs. complex objects
- note that the relationships between objects in
this example are more complex than hasPart or
isPartOf - because the model doesnt just deal with digital
objects - it may be worth drawing a distinction between
- compound objects (where objects have whole /
part type structural relationships) and - complex objects (where there are arbitrary
relationships between objects) ?? - most objects in digital libraries are complex
not just compound
16Summary why DC?
- DC (and the DCAM) provides a simple packaging
framework - where objects within the package are typically
passed by reference - highly flexible and extensible relationship
framework between objects - supports multiple syntax encodings
- compatible with Semantic Web (which allows for
possibility of inferencing across complex objects
from unknown sources) - content packaging is largely about relationships
i.e. it is just metadata
17Questions