This document provides an introduction to linked data for librarians. It discusses how computers currently lack understanding of meanings and relationships in metadata. It introduces the concept of representing all metadata as subject-predicate-object triples. It provides examples of representing information about goats using linked data. It discusses the five star model for open data and shows how the linked open data cloud has grown significantly. The document discusses how libraries can leverage their unique resources and apply linked data principles to improve discovery and provide open data for others. It acknowledges challenges but provides recommendations on where to start learning more about semantic web and linked data technologies.
An Introduction to Linked Data for Librarians (2018-06-28)
1. An Introduction to
Linked Data for Librarians
Cliff Landis
Special Libraries Association - Georgia Chapter
2018-06-28
2. The Metadata Problem: Computers are
stupid
(they don’t understand meanings or relationships)
• What we see • What the computer sees
<head></head>
<body>
<background>
<image>
<headline>Text Text Text</headline>
<tool></tool>
<text>Text Text Text</text>
</body>
https://www.auctr.edu/
7. Let’s make the Web a little smarter!
A young tree, shrub, vegetable,
or flower newly planted, or
intended for planting; a set, a
cutting, a seedling. Now chiefly
Eng. regional (midl. and south.)
and Irish English (north.): a young
cabbage plant. (from the OED)
=
Step 1: Disambiguation
10. Aiming for “5 star” Linked Open Data
● Available on the Web (whatever format) with an open license
(CC0, PD, ODC, RightsStatements.org)
● As above, but as machine-readable structured data (Excel
spreadsheets)
● As above, but in a non-proprietary format (CSV instead of
Excel)
● As above, but use URIs to denote things, so that people can
point at your stuff (persistent URIs, RDF, SPARQL)
● As above, but linking your data out to other data to provide
context (Linked Open Data)
http://5stardata.info/
11. Linked Open Data (2008-02-28)
https://lod-cloud.net/versions/2008-02-28/lod-cloud.png
12. Linked Open Data (2018-04-30)
"Linking Open Data cloud diagram 2018, by Andrejs Abele, John P. McCrae, Paul Buitelaar, Anja Jentzsch and Richard Cyganiak. http://lod-cloud.net/"
1,184 datasets with 15,993 links
13. Library of Congress Subject Headings
http://lod-cloud.net/clouds/lod-cloud.svg
19. Facts can be described with metadata
AUC Robert W. Woodruff Library has a website at https://www.auctr.edu/
20. Facts can be described with metadata
AUC Robert W. Woodruff Library has a website at https://www.auctr.edu/
AUC Robert W. Woodruff Library has a website https://www.auctr.edu/
21. Facts can be described with metadata
AUC Robert W. Woodruff Library has a website at https://www.auctr.edu/
AUC Robert W. Woodruff Library has a website https://www.auctr.edu/
SUBJECT PREDICATE OBJECT
22. Facts can be described with metadata
AUC Robert W. Woodruff Library has a website at https://www.auctr.edu/
AUC Robert W. Woodruff Library has a website https://www.auctr.edu/
SUBJECT PREDICATE OBJECT
http://dbpedia.org/page/Robert
_W._Woodruff_Library,_Atlanta
_University_Center
http://xmlns.com/
foaf/spec/#term_h
omepage
https://www.auctr.edu/
Robert W. Woodruff Library, Atlanta University Center Homepage
41. We’re sitting on a data goldmine
(but it has to be mined!)
•We can leverage our libraries’ unique content to
provide linked open data for others to remix and
reuse
•We can provide data we generate (statistics, analytics,
door count, etc.) as linked open data for us to mine,
and for others to mine nationally and globally
•We can use tools to generate information
visualizations of our own data and that of others
42. ...but I’m also a realist
• Linked open data is open. Once you publish data, you can’t control
how it will be used -- like walking in open spaces, you can’t control
who will incidentally photograph you
• Early adoption at large scales is often slow, difficult, and expensive
(but it’s finally starting to get easier and cheaper!)
• Managing linked open data is like any other metadata work -- it
requires work to setup, and regular maintenance to keep up
• Like all areas in the information economy, problems with quality
control and authority control (e.g. “fake news”) are migrating from
old media to new media -- requiring constant vigilance to ensure
you’re using trustworthy sources
43. Okay, okay, so now what?
Where to go from here
https://www.auctr.edu/
44. Semantic Web for beginners
WATCH
• Linked Open Data – What is it? – An introduction to LOD for memory organization
workers
• Tim Berners-Lee: The Next Web – Creator of the Web talks about its next stage of
evolution
• Linked Open Jazz – Using oral histories to explore relationships between artists
BROWSE
• WikiData.org – browse to get a feel for the subject-predicate-object relationships
• DBpedia.org – browse to get a feel for the use of LOD metadata standards
• lod-cloud.net - browse LOD datasets available for exploration and use
• SNAC Cooperative - browse how historical figures are connected through separate
archival collections
• LD4PE - browse Linked Data for Professional Educators to find learning resources