HiOA HiOA

Metadata interoperability

Nils Pharo, 01.10.2013

Overview

A bibliographic record in BIBSYS-MARC

*000 $a010041702
*008 $ap$bv$cnob$hno
*009c $ak
*015 $anf0105587
*020 $a82-05-27748-6$bib.
*080c $a839.6
*082g $d839.823[S]
*082kj$a839.82
*082xn$a839.82
*086d $aS 4b
*100 $aHamsun, Knut$d1859-1952
*245 $aSult$cKnut Hamsun
*260 $a[Oslo]$bGyldendal$c2001
*300 $a147 s.
*440 $aGyldendals 10 store
*500 $a1. utg. KÝbenhavn : Philipsen, 1890
*776 $w101353413

The bibliographic record converted to Dublin Core

DC.Format="p"
DC.Type="v"
DC.Language="nob"
DC.Identifier="82-05-27748-6"
DC.Subject="839.823[S]"
DC.Subject="839.82"
DC.Subject="839.82"
DC.Creator="Hamsun, Knut"
DC.Creator="1859-1952"
DC.Title="Sult"
DC.Publisher="Gyldendal"
DC.Date="2001"
DC.Description="1. utg. KÝbenhavn : Philipsen, 1890"
DC.Relation="101353413"

Mapping from BIBSYS-MARC to Dublin Core

Another MARC/DC example

Coen movie True Grit

Situation

Similar kind of content is described using different metadata standards and syntaxes

Why?

Create groups of four and discuss possible reasons why such situations occurr

Interoperability

"Interoperability is the ability of multiple systems with different
hardware and software platforms, data structures, and interfaces to
exchange data with minimal loss of content and functionality." (NISO)

 

Metadata interoperability

3 levels of interoperability (Chan & Zeng, 2006; Zeng & Chan, 2006)

Uniform solution - MARCsism

Everybody should use the same system! :-)

top down approach (MARCsism): "everybody should use MARC", is in theory the optimal solution, but is no longer realistic

Schema level interoperability

Derivation

To create a new schema from an existing source schema, e.g.:

The DC dumb down principle

Application profiles

Solutions based on a (combination of) existing schemas optimized for a specific community. The developers will typically have a bottom-up approach

Crosswalks

specifies the mapping of elements, semantics and syntax between different schemas. Very common way of securing interoperability. Two forms of crosswalks, absolute and relative

absolute crosswalk:
exact mapping between elements in the two schemas: MARC.260$c = DC.date.created
relative crosswalk:
mapping between elements that do not share eqivalent meaning:MARC.240 (Uniform title) = DC.title

Crosswalks cont.

Crosswalk examples

Switching across

To map between multiple schemas one of the schemas are used as a switch

Example: Getty's crosswalk where CDWA (Categories for the Description of Works of Art) is used to map between 12 different schemas

Metadata frameworks

Created to provide guidelines for developers of metadata schemas in specific environments. Frameworks can be developed based on existing schemas or prior to any schema has been developed

Example: OAIS reference model

Metadata registries

Provide overview of relevant metadata schemas, their elements, syntax, semantics etc to facilitate adoption and reuse of existing schemas in favour of creating new (and redundant) schemas

Example: Dublin Core Metadata Registry

Record level interoperability

when it's too late for schema level solutions

Metadata conversion

convert the content of metadata records in one schema into another

Data reuse and integration

This approach is based on combining metadata from various sources in one common format. METS (Metadata Encoding and Transmission Standard) and RDF provide two different solutions for this.

RDF example

<?xml version="1.0"?>
<rdf:RDF
xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:dc="http://purl.org/dc/elements/1.0/"
xmlns:skos="http://www.w3.org/2004/02/skos/core#">
<rdf:Description rdf:about="http://mitpress.mit.edu/catalog/item/default.asp?ttype=2&tid=3792">
<dc:title>The Intellectual Foundation of Information Organization</dc:title>
<dc:creator>Elaine Svenonius</dc:creator> <dc:publisher>MIT Press</dc:publisher>
<dc:date>1999</dc:date>
<dc:language>en</dc:language>
<skos:prefLabel>information organization;/skos:prefLabel>
</rdf:Description>
</rdf:RDF>

Repository level interoperability

Metadata repositories that collect metadata from many providers, how do they deal with interoperability issues?

Different solutions:

No conversion

Aggregation of content

The content of a metadata record for an item is the aggregated contents of all the item's metadata records,(NSDL , DLESE)

aggregation of metadata records

Repository crosswalks

To ensure consistency in multi-collection databases

Elements or values can be the subject of crosswalking

Reality check

Norwegian example, two repositories and one harvester: