It19 20140721 linked data personal perspective

Janifer Gatenby

OCLC EMEA

With acknowledgements to Richard Wallis and Anila Angjeli

LINKED DATAA PERSONAL PERSPECTIVE

LINKED DATA

• What is it?

• What does it promise?

• How do we get there?

• What happens when we get there?

WHAT IS IT?

A WAY OF EXPRESSING A LINK

What is it?

• Not really a new way of linking but a new way of expressing a link

It is about using canonical trusted globally

referenceable identifiers for concepts, people,

organisations, locations etc. instead of copying text

strings and losing the connection with the

authoritative sources they came from.

Richard Wallis

MARC21 LINKS

What is it?

• 700 10 $a name $e role $0 authority control number

– (added entry in a MARC record for a name related to a work, not the main

author)

These familiar links reference an authority record in the

same database as a bibliographic record, hence have

no address portion. Linked data extends the linking

range.

EXTENDING THE LINKING RANGE: URI

What is it?

• URI – immutable address as well as an identifier

– http://id.loc.gov/authorities/names/nr89009099

– http://viaf.org/viaf /116774723

– http://isni.org/isni/000000114556841

9 NACO libraries –

LC,

National Agricultural Library,

National Library of Medicine,

British Library,

NL Mexico,

NLNZ,

NL Scotland,

NL South Africa,

NL Wales

• RDF – metadata is expressed in triples

– Data

– Data label (properties)

– Vocabulary from which the label comes (gives context to the label)

EXTENDING THE LINKING RANGE: RDF

What is it?

SPARQL

What is it?

• A database can offer a SPARQL endpoint = can receive RDF queries

– Author [schema] Name [data label] De Groot, Gerard J., 1955 [data]

• “SPARQL allows users to write queries against data that can loosely be called "key-value" data,

more specifically it is data that follows the RDF specification of the W3C. The entire database is

thus a set of "subject-predicate-object" triples.”

• 1.1 Stable release 2013-03-21

– W3C recommendation

http://en.wikipedia.org/wiki/SPARQL

http://www.w3.org/blog/SW/2008/01/15/sparql

_is_a_recommendation/

LINKED DATA PRINCIPLES

What is it?

1. Use URIs as names for things

2. Use HTTP URIs so people can look up those names

3. When someone looks up a URI, provide useful information, using the standards - RDF

4. Include links to other URIs, so that they can discover more

Tim Berners-Lee - 2006

VOCABULARIES

What is it?

• Vocabularies are not schemas, they are lists of defined data labels (concepts)

– Schema.org (Search engines)

– BibFrame (Library community)

– FOAF Friend of a friend

– OWL same as

• Vocabularies can be mixed foaf:name "Jimmy Wales" ;

foaf:mbox <mailto:jwales@bomis.com> ;

foaf:homepage <http://www.jimmywales.com/> ;

foaf:nick "Jimbo" ;

WHAT DOES IT PROMISE?

What does it promise?

• Enriched displays without data maintenance

• Better harvesting and ranking

• because of markup

• and because of links

• Navigation to pages with additional information –

– Example: from VIAF via ISNI to encyclopaedias, rights management societies (digitisation

rights), Bowker – biographies from fly leaves

INTERCONNECTING FRENCH CULTURAL HERITAGE TREASURES ON

THE WEB

What does it promise?

BnF Main catalogue

(MARC)

Digital documents

(DC)

Web pages for

Internet usersBnF Archives and

Manuscripts

catalogue

(EAD) Raw data for machines

Modeling

Matching

Clustering

Alignments

Semantic Web

techniques

Other BnF

resourcesExternal

resources

What does it promise?

BnF persistent ID

Imported

from

Wikipedia

and

integrated in

the page

What does it promise?

Information about the data model (or ontology) at : http://data.bnf.fr/about-en

Data can be downloaded

Existing ones + othersdefined for the specific

needs of the project

BIG DATA AS RDF

• Data is re-usable without a full blown conversion

• Permits 3rd party analysis of big data sets

• Data mining for new information

What does it promise?

HOW DO WE GET THERE?

MAKING THE LINKS

How do we get there?

DNB CultureGraph

– “It’s all about creating

connections”

– DDC to RVK (German

classification) by

comparing search

results

– GND (names) to

German Wikipedia

EXAMPLE: VIAF

How do we get there?

• Ingesting data to compare and create links

• Makes clusters; cluster identifier

• Ingesting preferred to external linking

– Wikipedia, ISNI, WorldCat identities

– More data used for clustering, so more reliable

• VIAFBot for making reciprocal links in Wikipedia / Wikidata

<rdf:type rdf:resource="http://xmlns.com/foaf/0.1/Person"/>

<rdf:typedf:resource="http://rdvocab.info/uri/schema/FRBRentitiesRDA/Person"/>

<foaf:name>De Groot, Gerard J., 1955-</foaf:name>

<foaf:name>DeGroot, Gerard J., 1955-</foaf:name>

<rdaGr2:dateOfBirth>1955-06-22</rdaGr2:dateOfBirth>

<owl:sameAs rdf:resource="http://data.bnf.fr/ark:/12148/cb12299846b#foaf:Person"/>

<owl:sameAs rdf:resource="http://www.idref.fr/034977651/id"/>

<owl:sameAs rdf:resource="http://d-nb.info/gnd/12422900X"/>

Libraries

Text Rights

Music RightsTrade Sources

Encyclopaedias

Researchers & ProfessionalGranting organisationsProfessional SocietiesArticle databasesTheses databases

cross-domain bridging-domains

Archives and Museums

EXAMPLE: ISNI: 15 MILLION LINKS

How do we get there?

Linked Data: isni.org/isni/

LA TROBE UNIVERSITY LINKS: 3,427

How do we get there?

LA TROBE UNIVERSITY: 1,864 VIAF LINKS

How do we get there?

ISNI – A LINKING IDENTIFIER

How do we get there?

• Identifiers Seal Uniqueness: “n” number

of other elements are necessary for

uniqueness

• Stable identifier; stable metadata:

• assigned where there is confidence in

the quality and completeness of the

metadata to establish uniqueness

• ISNI system + Quality Team (BL & BnF)

Linking erroneous data

propagates errors.

LINKS ARE MADE ONCE – THEN INHERITED

How do we get there?

• URI – immutable address as well as an identifier

– http://id.loc.gov/authorities/names/nr89009099

– http://viaf.org/viaf /116774723

– http://isni-url.oclc.nl/isni/000000114556841

9 NACO libraries –

Library of Congress,

National Agricultural Library,

National Library of Medicine,

British Library,

NL Mexico,

NLNZ,

NL Scotland,

NL South Africa,

NL Wales

WHAT HAPPENS WHEN WE GET THERE?

HOW DOES SEARCHING WORK?

• Search happens mostly in the search engines

• Library catalogue concentrates on:

– Being linked to

– Linking out (navigation)

– Delivery, particularly of the digitised and immediate

What happens when we get there?

HOW DO SEARCH AND LINKED DATA INTERACT?

• Is search really fully delegated to search

engines & larger union catalogues?

What happens when we get there?

SEARCH TYPES

What happens when we get there?

Search type Happening in

Known item Search engines, also in more specific

sources where noise is a problem

Subject search Search engines, also in more specific

sources, to reduce noise and benefit from

more precise searching capabilities

Index browse In catalogues

Follow a link Everywhere . In library catalogues from a

full record display.

The more your catalogue is linked in, the more likely it is

to attract all types of searches

STORE ONLY THE LINKS?

What happens when we get there?

• Data needed

• For making indexes

• For comparisons,

e.g. For de-

duplication

• Data mining

It is about using canonical trusted globally

referenceable identifiers for concepts, people,

organisations, locations etc. instead of copying text

strings and losing the connection with the

authoritative sources they came from.

This doesn’t mean that you only

need the links; you often also

need to ingest the data

Besides data storage no longer the constraint it once was

READ FURTHER

• http://www.slideshare.net/tulipbiru64/the-single-power-of-link-richard-wallis

• http://www.slideshare.net/rjw/linked-data-and-oclc