Linked Data as an enabling framework for resource discovery across libraries, museums and archives
-
Upload
andy-powell -
Category
Technology
-
view
2.623 -
download
0
description
Transcript of Linked Data as an enabling framework for resource discovery across libraries, museums and archives
www.eduserv.org.uk
Linked Data as an enabling framework for resource discovery across libraries,
museums and archives
Andy Powell (and Pete Johnston)Eduserv
@andypowe11
www.eduserv.org.uk
Image by: ...-Wink-... @ Flickr
www.eduserv.org.uk
What’s coming…
• report on some work Pete Johnston and I (both at Eduserv) undertook in March 2011
• on behalf of the JISC and RLUK Resource Discovery Taskforce
• (which subsequently became “Discovery”)• to develop some metadata guidelines for use
across libraries, museums, and archives
http://bit.ly/hmMvP1
www.eduserv.org.uk
Functional requirement
• help libraries, museums and archives expose existing metadata (and new metadata created using existing practice) in ways that– support the development of aggregator services– integrate well with the web (and the emerging
web of data)• note: NOT re-engineering cataloguing practice
in the LAM sectors
www.eduserv.org.uk
Guiding principles
• support the RDTF Vision• in line with Linked Data principles• based on the W3C Linked Open Data Star Scheme• in line with Designing URI Sets for the UK Public
Sector• take into account the Europeana Data Model and ESE• be broadly in line with the notion of “making better
websites” across libraries, museums and archives
www.eduserv.org.uk
RDTF Vision
“making the most of UK scholarly and cultural resources by best positioning their metadata for
discovery and reuse within the global information ecosystem”
www.eduserv.org.uk
Linked Data
• use URIs as names for things• use HTTP URIs so that people can look up
those names• when someone looks up a URI, provide useful
information, using the standards (RDF, SPARQL)
• include links to other URIs so that they can discover more things
http://www.w3.org/DesignIssues/LinkedData.html
www.eduserv.org.uk
Linked Open Data Star Scheme
• 1 star - make your stuff available on the Web (whatever format) under an open license
• 2 star - make it available as structured data (e.g., Excel instead of image scan of a table)
• 3 star - use non-proprietary formats (e.g., CSV instead of Excel)
• 4 star - use URIs to identify things, so that people can point at your stuff
• 5 star - link your data to other data to provide context
http://bit.ly/u1O7e3
www.eduserv.org.uk
Our draft RDTF proposal
• used the W3C Linked Open Data star scheme as framework (at 3, 4 and 5 star levels)
• and suggested three approaches– community formats– RDF data– Linked Data
• 196 comments – on pretty much all aspects of the draft
www.eduserv.org.uk
Re-conceptualising the guidelines
Collectionsof Descriptions
RDFNot-RDF
Individual ItemDescriptions
www.eduserv.org.uk
The draft guidelines
Collectionsof Descriptions
“RDF Data”“bulk transfer”
RDFNot-RDF
Individual ItemDescriptions
Linked Data
www.eduserv.org.uk
The Web!
Collectionsof Descriptions
“RDF Data”“bulk transfer”
RDFNot-RDF
Individual ItemDescriptions
Linked Data“page per thing”
www.eduserv.org.uk
Guiding principles
• support the RDTF Vision• in line with Linked Data principles• based on the W3C Linked Open Data Star Scheme• in line with Designing URI Sets for the UK Public
Sector• take into account the Europeana Data Model and ESE• be broadly in line with the notion of “making better
websites” across libraries, museums and archives
our i
nitia
l dra
ft p
roba
bly
faile
d in
this
!!
www.eduserv.org.uk
Semantics vs. linking
• moving left to right across the quadrants…– offers greater semantic precision within a more
consistent framework (RDF)
“RDF Data”“bulk transfer”
Linked Data“page per thing”
www.eduserv.org.uk
Linking vs. semantics
• moving bottom to top across the quadrants…– promotes the individual descriptions (rather than
collections of descriptions) and encourages the assignment of identifiers (i.e. URIs) to both those descriptions and the things they identify
“RDF Data”“bulk transfer”
Linked Data“page per thing”
www.eduserv.org.uk
Possible adoption path
Collectionsof Descriptions
“RDF Data”“bulk transfer”
RDFNot-RDF
Individual ItemDescriptions
Linked Data“page per thing”
www.eduserv.org.uk
Bulk transfer
• “give us what you’ve got”• serve existing community bulk-formats (e.g. files
containing collections of MARC, MODS, BibTeX, DC/XML, SPECTRUM or EAD records) or CSV over RESTful HTTP
• use sitemaps and robots.txt and/or RSS/Atom to advertise availability and GZip for compression
• for CSV, provide a column called ‘label’ or ‘title’ so we’ve got something to display
• give us separate records (for CSV, read ‘rows’) about separate resources (where you can)
“RDF Data”
“bulk transfer”
Linked Data
“page per
thing”
www.eduserv.org.uk
Page per thing
• “build better websites”• serve an HTML page (i.e. a description) for every “thing” of
interest over RESTful HTTP• optionally serve alternative format(s) for each description
(e.g. a MODS or DC/XML record) at separate URIs and link from the HTML descriptions using “<link rel=“alternative” … />
• use “cool” ‘http’ URIs for all descriptions• use sitemaps and robots.txt and/or RSS/Atom to advertise
availability• optionally offer OAI-PMH server to allow harvesting
“RDF Data”
“bulk transfer”
Linked Data
“page per
thing”
www.eduserv.org.uk
RDF data
• “RDF bulk download”• serve big buckets of RDF (as RDF/XML, N-Tuples or N-
Quads) over RESTful HTTP• re-use existing conceptual models and vocabularies
where you can• assign URIs to every “thing” of interest• use Semantic Sitemaps and the Vocabulary of
Interlinked Datasets (VoID) to advertise availability of the buckets
“RDF Data”
“bulk transfer”
Linked Data
“page per
thing”
www.eduserv.org.uk
Linked Data
• “W3C 5 star approach”• serve HTML and RDF/RDFa for every “thing” of
interest over RESTful HTTP• assign ‘http’ URIs to every “thing” (and every
description of a thing)• follow “cool URIs for the semantic web”
recommended practice• become part of the web of data - link to other
people’s stuff using their URIs
“RDF Data”
“bulk transfer”
Linked Data
“page per
thing”
www.eduserv.org.uk
Possible adoption path
Collectionsof Descriptions
“RDF Data”“bulk transfer”
RDFNot-RDF
Individual ItemDescriptions
Linked Data“page per thing”
www.eduserv.org.uk
Where are we headed?
• implication of previous slide is that “5 star” Linked Data is where we want to be
• some good reasons for that– rich semantic framework– “follow your nose” approach to getting more info– “small pieces loosely joined”– link and be linked – data becomes “of” the web– relatively easy re-use of other peoples’ ontologies
www.eduserv.org.uk
But…
• also need to remember that “the best is sometimes the enemy of the good”
• recent web history tells us that uptake of complex technologies can be a slow process!
• certainly been the case with the Semantic Web and RDF
www.eduserv.org.uk
‘Linked’ and ‘social’ can be a win
• Open Graph Protocol• as proposed by Facebook but now more
widely adopted• good example of Linked Data underpinning
social activity (part of which is related to discovery)
• but what’s more important – the fact it uses RDF or the fact it uses the HTTP URI?
http://ogp.me/
www.eduserv.org.uk
I would suggest the latter
www.eduserv.org.uk
Conclusions
• true Linked Data is a good aim for libraries, museums and archives…
• …but our emphasis should be on the ‘linked’ in short term
• encouraging more item-level material on the Web with cool URIs (even if only in largely human-readable form)
• richer Linked Data can then emerge over time
www.eduserv.org.ukwww.eduserv.org.uk
Image by: ...-Wink-... @ Flickr
Questions?