1. A Registry for controlled vocabularies at the Library of Congress
Rebecca Guenther
Network Development & MARC Standards Office,
Library of Congress
October 29, 2008

2. Outline of presentation
Types of controlled vocabularies
Vocabularies maintained at LC
An introduction to SKOS
Establishing concept databases at LC
Examples of concept schemes: ISO and PREMIS event type
Providing the registry as a web service
ASIST 2008
Oct. 29, 2008

3. Why establish controlled vocabularies?
Control values that occur in metadata
Document and publish for reuse
Reduce ambiguity
Control synonyms
Establish formal relationships among terms (where appropriate)
Test and validate terms
Many metadata schemes allow for content from other sources. Some data elements may be more useful if a controlled vocabulary is used. Some are published formally, others are developed and used locally.
Formal controlled vocabularies may be used for testing and validation of terms– this is often done in integrated library systems, where bibliographic records may validate against authority records. This is one instance of testing and validation of terms. There is work being done on establishing metadata registries for both documentation and machine validation of both controlled vocabularies and metadata elements/terms. This could be particularly useful for controlled vocabularies, since their usefulness depends on consistency.
ASIST 2008
Oct. 29, 2008

4. Types of Controlled Vocabularies used in metadata standards
Lists of enumerated values
Code lists (e.g. language, country)
Taxonomies
Formal Thesauri
Locally controlled enumerated lists
NISO has a standard for constructing thesauri (free download – in bibliography)
ANSI/NISO Z : Guidelines for the Construction, Format, and Management of Monolingual Controlled Vocabularies
ASIST 2008
Oct. 29, 2008

5. Enumerated lists
Simple list of terms used in a pull-down menu or Web site pick list
Values enumerated in an XML schema
Little additional information or structure about each value
Examples:
Code and value from a MARC 21 fixed field, e.g. code “e” in Leader/06 is “cartographic material”
Enumerated value “MD5” for METS CHECKSUMTYPE
Enumerated value “born digital” in MODS digitalOrigin
ASIST 2008
Oct. 29, 2008

6. Code lists
Some established as ISO standards and used worldwide in many communities for many purposes
The standard standardizes the code, not a particular name for it
Codes are used as identifiers
Examples (maintained by LC):
ISO (language codes)
MARC relator codes
MARC country codes
ASIST 2008
Oct. 29, 2008

7. Thesauri
A thesaurus is a controlled vocabulary with multiple types of relationships
Example:
Rice
UF paddy
BT Cereals
BT Plant products
NT Brown rice
RT Rice straw
ASIST 2008
Oct. 29, 2008

8. Standards maintained at LC that use controlled vocabularies
MARC (including code lists)
MODS
METS
MIX (XML schema for Z39.87 Technical metadata for digital still images)
PREMIS
ISO (language codes)
Thesaurus of Graphic Materials
LCSH
… and some others
ASIST 2008
Oct. 29, 2008

9. Simple Knowledge Organisation System(s)
SKOS: What is it?
Simple Knowledge Organisation System(s)
SKOS is …
for declaring and publishing taxonomies, thesauri or classification schemes, for use in a distributed, decentralised information system (i.e. a semantic web).
for describing Concepts and creating relationships between Concepts and Terms
A practical application of RDF
a formal language for representing controlled, structured vocabularies
ASIST 2008
Oct. 29, 2008 9

10. The SKOS data model
…views a knowledge organization system as a concept scheme comprising a set of conceptual resources (concepts).
These concept schemes and conceptual resources are identified by URIs.
The model is multilingual and extensible
ASIST 2008 10
Oct. 29, 2008

11. Concepts can be…
labeled with any number of strings. One label, in any given language, can be indicated as the "preferred" label for that language, and others as "alternate“ labels, "hidden“ labels, or using a notation:
skos:prefLabel
skos:altLabel
skos:hiddenLabel
skos:notation
ASIST 2008 11
Oct. 29, 2008

12. Concepts can be…
linked to other concepts within the same concept scheme.
Hierarchical links:
skos:broader and skos:narrower
skos:broaderTransitive and skos:narrowerTransitive
Associative links:
skos:related
ASIST 2008 12
Oct. 29, 2008

13. Concepts can be…
grouped into collections, which can be labeled and/or ordered. A concept can be in one or more collections
skos: Collection
skos: OrderedCollection
skos: member
skos: memberList
ASIST 2008 13
Oct. 29, 2008

14. Concepts can be…
mapped to other concepts in different concept schemes.
Hierarchical mapping:
skos:broadMatch
skos:narrowMatch
Associative mapping:
skos:relatedMatch
skos:closeMatch
skos:exactMatch
ASIST 2008 14
Oct. 29, 2008

15. Advantages to using SKOS
SKOS has a defined element set which is particularly relevant for controlled vocabularies
Relationships between entries in a thesaurus can be expressed (broader, narrower, etc.)
Relationships between entries in different thesauri can be expressed (exactMatch, related)
Having a dereferencable URI for concepts and their concept schemes enhances the ability to provide web services for consumers of these standards
ASIST 2008
Oct. 29, 2008

16. Controlled vocabularies registry at LC
Library of Congress is establishing databases with controlled vocabulary values for standards that it maintains
Controlled lists are represented using SKOS as well as alternative syntaxes
Lists currently in progress:
ISO and MARC language code list
MARC geographic area codes
MARC country code list
MARC relators
PREMIS controlled value lists
Thesaurus of Graphic Materials
Other possibilities
Enumerated values in MODS schema
Coded and uncoded value lists in MARC
ASIST 2008
Oct. 29, 2008

17. Reasons for developing a registry
Facilitate development and maintenance process
Make controlled lists openly available
Develop a web service where comprehensive information about controlled terms is available
Experiment with semantic web technologies
Expose vocabularies to a wider communities
ASIST 2008
Oct. 29, 2008

19. Example: ISO 639-2 vocabulary
One in the family of ISO 639 language coding standards
Has a close relationship with other language coding standards (ISO and -3, MARC)
LC is maintenance agency
The standard is the CODE, not the language name; multiple names are given
ASIST 2008
Oct. 29, 2008

21. ISO 639-2 language code example
<rdf:Description rdf:about= "
<rdf:type rdf:resource=" #Concept"/>
<skos:prefLabel xml:lang="x-notation">por</skos:prefLabel>
<skos:altLabel xml:lang="en-Latn">Portuguese</skos:altLabel>
<skos:altLabel xml:lang="fr-Latn">portugais</skos:altLabel>
<skos:notation rdf:datatype="xs:string">por</skos:notation>
<skos:definition xml:lang="en-Latn">This Concept has not yet been defined.</skos:definition>
<skos:inScheme rdf:resource="
<vs:term_status>stable</vs:term_status>
<skos:historyNote rdf:datatype="xs:dateTime"> T08:41: :00</skos:historyNote>
<skos:exactMatch rdf:resource= "
<skos:changeNote rdf:datatype="xs:dateTime"> T13:49: :00</skos:changeNote>
</rdf:Description>
This is one type of SKOS expression, which is RDF/XML. Tags defined by SKOS a wrapped in an RDF wrapper. skos:prefLabel uses the language code as value so as to not give a preference to a term in a particular language. The skos:altLabel is used here for the various language names
skos:inScheme tells you what concept scheme this entry is included in; there can be multiples.
skos:exactMatch gives a URI for the other code which is exactly the same; in this case there is the 2-character code “pt” (ISO 639-1), which is equivalent to this one.

22. PREMIS controlled lists
PREMIS Data Dictionary for Preservation Metadata
Some semantic units call for controlled vocabularies and have suggested lists
A central registry could document and make them available
Users could submit their own terms
PREMIS schema could be enhanced with enumerated values for validation generated dynamically
ASIST 2008
Oct. 29, 2008

25. PREMIS event type example
<rdf:Description rdf:about= "
<rdf:type rdf:resource= "
<skos:prefLabel xml:lang="en-latn"> creation</skos:prefLabel>
<skos:narrower rdf:resource= "
<skos:narrower rdf:resource= "
<skos:definition xml:lang= "en-latn">the act of creating a new object</skos:definition>
<skos:inScheme rdf:resource= " /preservationEvents"/>
</rdf:Description>
This example is from a concept scheme called “preservation events”. The controlled value described is “creation”, which is in the PREMIS data dictionary as a suggested value under eventType. It is a broader term for 2 others on the value list: migration and normalization.

26. XML Database using XQuery
Registry Web service
XML Database using XQuery
(eXist)
RDF Triple Store
(Sesame)
HTTP request
User
Runs query Gets results Sends back to database and then to user
Interprets URI Formulates SPARQL query

27. Further development Consider programming changes to improve speed
Develop mechanisms to output all public documentation from database
Include additional coding about relationships to other concept schemes and controlled vocabularies (facilitating crosswalks)
Encourage experimentation
ASIST 2008
Oct. 29, 2008

28. Questions? Contacts: Rebecca Guenther: rgue@loc.gov
Clay Redding:
ASIST 2008
Oct. 29, 2008