Upload
jack-gibbs
View
214
Download
1
Embed Size (px)
Citation preview
New Century, New Metadata
Thomas Krichelhttp://openlib.org/home/krichel
University of Surrey, Hitotsubashi University and Long Island University
Why Metadata
FunInformation retrievalSupport organization of social process
Crisis of Author Self-archiving
Formal archivingSmallMetadata poor
Informal archivingInformation retrieval difficultLack of support infrastructure
Improving formal archiving
Strengthen the metadata provisionBroaden the mission of archivingAllow usage of archived material in many
user servicesBetter report on archive material usageStrengthen the relationship with overlay
services
Improving Informal Archiving
Build standardized metadata supply format
Harvest that metadata into larger digital libraries
Offer archival backup for papers
Metadata to Support Self-archiving
Simple to composeIntuitive vocabulary that is specific to the
academic process, e.g. “author” instead of “creator”
Widely applicableAll disciplines and publication forms
High quality i.e. controlled
Metadata Control
Any processing that is done to the metadata before its inclusion in a user service.
Essential in a situation where metadata is harvested.
Types of Control
Syntactic controlRelational controlRetrieval controlIdentity controlVerity controlAccession control
Basic Model
Four different record typesDocumentGroupPersonOrganization
Group and document
There is only one document type.Groups are used to refine the status of
the document.Group construct meant to be defined by
librarians, publishers and other intermediaries.
Person and Institution
Person and institution admit very similar attributes
It is hoped that organizational information will be contributed by intermediaries.
Implementation of Basic Model
RePEc100000 documents100 groups (series)500 authors5000 institutions
Examplehttp://ideas.uqam.ca/EDIRC/data/frbgvus.html
Possible to do the same thing for ReLIS
Basic Grammar
XML syntaxThree groups of XML elements
Nouns: element for items describedAdjectives: elements that describe nounsVerbs: elements that relate nouns
Modular Design
<person><isauthorof>
<document><ispublishedby>
<organization><hasmember>
<person></person>
</hasmember></organization>
</ispublishedby></document>
</isauthorof></person>
Relational Design
<person id=“kmarxthered”><email> [email protected]</email> </person>
<document id=“kapital”> <title>Das Kapital</title><hasauthor> <person id=“kmarxthered”/> </hasauthor></document>
Other features
Lang qualifier to all elements, it ISO 639-1 if there are two letters and the bibliographic variant of ISO 639-2 if three letters.
Nouns have id.Verbs have startdate and enddate
qualifiers, and of course have id.Adjectives can have child elements.
Remaining Problems
Resolvability rules for identifiersDates and historySubject classification using the group
mechanismAliasing of element names
To be done…
Complete list of verbs and adjectivesSchema designParsing and validation software.Conversion with test collection ReLIS.
Collaboration is welcome
Thanks for listening.
Have a happy New Year.