FSR 2014, Rome Which knowledge organization systems for

Download Report

Transcript FSR 2014, Rome Which knowledge organization systems for

FSR, 17-18 Feb 2014, Rome
Which
knowledge organization systems
for conceptual interoperability?
Claudio Gnoli
ISKO Italy
Convergence
Libraries (IFLA)
Archives (ICA, IASA)
Museums (ICOM)
+
Galleries
Zoological gardens
Botanical gardens
Websites
etc.
[University of Virginia
Art Museum on Vimeo]
All store and give access to
documents
(information resources)
Convergence
Encouraged by the digital formats
[Rayward as early as 1998]
Ideally, people should input a single search
and be able to navigate between digital documents
(or digital representations of material documents)
irrespective of their nature, source and institution
Convergence
...but is all this real ??
Interoperability
The Internet, XML-RDF, linked data
are boosting technical interoperability.
What about conceptual interoperability?
Interoperability
<dc:Subject>590</dc:Subject>
<owl:Class>Animaux</owl:Class>
Technically but not conceptually
interoperable metadata
Several barriers implied here...
Potential information sources
Example:
“where can I find information on
badgers in Tuscany?”
Potential information sources
Well,
in books...
Potential information sources
… in adademic
papers ...
Potential information sources
...in local
museums...
[Wikimedia]
Potential information sources
...in zoological
gardens
[dailymash.co.uk]
Potential information sources
(sometimes with
Twitter output) ...
Potential information sources
… or even in churches
fresco in the abbey of
Monte Oliveto Maggiore
(Tuscany)
Potential information sources
All these sources can provide
some relevant information
(e.g., that there were tame badgers
in 16th c. Tuscany)
So where is their union catalogue ?...
KOS
The only bridge
to retrieve different
information sources
on a same subject
are
knowledge organization
systems (KOS)
KOS
The notion of KOS is useful as it includes
tools developed in different communities:








keywords and tags
terminologies
taxonomies
subject heading lists
} controlled
thesauri
}
vocabularies s.s.
classification schemes
ontologies
...
Different features but common basic principles
Knowledge Organization (KO)
A term in use since the 1970s [Dahlberg]
with precursors [Bliss 1929]
includes “subject indexing”,
“classification”,
“categorization”,
“book indexing”,
“taxonomies”,
“system of the sciences”...
Limits to interoperability
Acknowledging the common notion of KOS
is a first step.
However, there are many different KOSs
and KOS types
which limits conceptual interoperability in several ways:
(1) language
(2) disciplines
(3) representation
(4) alignment
Limits to interoperability.
1: Language
All verbal KOSs are only useful
to speakers of their language.
en:
fr:
it:
de:
ru:
zh:
badger
blaireau
tasso
Dachs
yazvik
獾亚科
Limits to interoperability. 1
We need multilingual verbal KOSs (e.g. MACS)
or classification systems based on notation (e.g. UDC)
Limits to interoperability.
2: Disciplines
Library classifications are based on disciplines,
while taxonomies or thesauri are not.
Please, are you husbandry,
or zoology,
or entertainment?...
Well, I'm a badger...
Limits to interoperability.
2: Disciplines
The León Manifesto recommends that KOSs
distinguish phenomena from disciplines and carriers
and allow to search independently for each, e.g.:
horses
horses
badgers
badgers
badgers
in
in
in
in
in
military science
husbandry
husbandry
comics
frescos
Limits to interoperability.
3: Representation
Markup formats don't always represent
all features of a KOS. E.g.,
full faceted structures cannot be represented
in MARC (based on non-faceted DDC and LCC),
nor in SKOS (based on thesauri) [Gnoli et al. 2011].
<Growth> skos:facetOf <Animals>
<Growth> skos:category <Process>
Limits to interoperability.
3: Representation
Most CMSs for the Web
don't manage
systematic sorting
through a notation.
OWL allows for
any relationship [Zeng et al. 2010],
but is mostly used for hierarchical ones
until now...
Limits to interoperability.
4: Alignment
Every KOS organizes knowledge in its own way.
Mapping is not always obvious:
Aubergines
Horticulture
=EQ Egg-plants
~EQ Gardening
Inland waterways EQ Rivers|Canals
[Dextre Clarke 2011 on ISO 25964]
Limits to interoperability.
4: Alignment
According to domain analytical theory [Hjørland],
KOSs may be incommensurable.
Others are more optimistic [Szostak]:
circumlocutions always helped translation and trade
Fuzzy relationships as in ISO 25964
may be a solution.
Still, mapping is costly...
[sheppardsoftware.com]
Recommendations
KOSs should be interoperable as for:
(1) language:
<... lang=“en”>
(2) disciplines
<Animals> studiedBy <Zoology>
(3) representation Publish (your KOS in LOD) or perish
(4) alignment
Invest in mapping
Recommendations
– Use KOSs! (Any is better than none.)
- Use interoperable KOSs!
Thanks for your attention!
[email protected]
@scritur