SWAD-Europe Initial workshop - rough notes taken by Libby Miller, minimally edited by Charles McCathieNevile
Attendees
Tom Baker
Dublin Core
Sandy...
Librarian
Boris Lauser
Food and Agriculture Organisation (United
Nations)
Gauri Salokhe?
Food and Agriculture Organisation (United
Nations)
Libby Miller
ILRT
developer, query languages for RDF
Stefan
German University Library
Grant campbell
University of Western Ontario
to undertand practical tools for incorporating RDF and sw into grad
and undergrad studies
Susanna Peruginelli?
previously from Florence library; now freelance consultant
Doug Campbell
National Library of New Zealand
Developer
??
Italian parliament
??
member of national metdata working group, Rome
?Ana Baptista
Universidad do Minho
IS dept
RDF in research dept
?Maewyn Cumming
E-envoy's office, UK government
Stefan?? female, german, library
distributed schema translation, app profiles
Traugott Koch
encoding vocabulary schemes, registry context
Harry Wagner
developer OCLC
Liddy Nevile
understanding how RDF useful for us, accessibility
Eric Childress
from OCLC - a standard publication format - publication
Charles McCathieNevile
builds small tools
things can build easily to see how it works what can I do to help
me about cataloging information all the time
Jens Vindvad?
Norwegian Libraries
Roberto Dallari
Library of the Italian Parliament
Simon Pockley
Australian Centre for the Moving Image
built own tools, xml
need to ransack other peoples' catalogues - who has the best print
for film. need to exchange metadta
Joe Tennis
University of Washington
classification theory research
interest in controlled vocab intereoperability
Haruki Nagata?
from Japan
now University of Tsukuba - inetenationalization and registry
XML suffices sometimes; RDF others - when should you use one and when the other? (Andy Powell)
How are RDF and XML related?
How do topicmaps relate to RDF?
Scalability and performance of RDF (Harry Wagner)
What tools and apps are avilable for end users? (danbri)
Easy introduction documents, tutorials? (danbri)
How do we merge and share data when we don't always identify things in the same way - in practice, uris opoen dont exist or people dont know what they are. (handling anonymous resources and resources with multiple identifiers)
Access control - how do you protect information? (we are exposing our information to others' systems - people in want to have security - don't want to share, or don't want to share wrong thing.) (danbri: close to creative commons - give people access but say what they can use it for.)
Who should we contact to find solutions to problems like: -
annotating content (e.g. annotating aboriginal work) - user end
tools - statements in RDF about accessibility - can this user agent
access this content as it is, or not until its been transformed in
some way (e.g. annotating the stylesheet).
(Liddy)
How can we handle provenance tracking (e.g. 5 people making 5
different claims about a page's accessibility)
(Charles)
How should one associate metadata with different objects? (Eric Miller)
Do you have any advice about distinguishing between the object and a manifestation of the object? This keeps cropping up (Simon)
How can you extract RDF from legacy systems?
How do I/should I use XML databases for XML/RDF content? How good is 'native' RDF database performance? What are the advantages of different sorts of databases? What tools are available to manipulate the metadta in its various locations? (Simon)
What is the appropriate level of detail of description for an object? (Danbri)
At what level of granularity should I describe my resources? e.g. how much do you encode the data? date, gregorian month, gregorian day etc (Charles McCathieNevile)
If people use different levels of granularity, how can we link them together? (danbri)
Is there a role for XML and RDF in bringing ERP systems, doc mangement systems, desktops together - getting at the information through one interface? 'enterprise data integration' (Sandy)
How should you handle versioning for XML schemas and DTDs and RDF schemas? What are the pros, cons and implications of different approaches? (Eric Miller)
What is the state of the art in tools and best practice for multilingual
XML/RDF?
(Charles) [see discussion of this
question below - CMN]
Where do I put the meta data?
(Charles)
A:
danbri - in html docs, creates links to pure rdf with links to
other rdf other times it is appropriate to embed in media, e.g.
XMP, jpg
charles - for documents others have created, cant change that document, so he uses Annotea
harry - good to have descriptions of why people do it in different ways
---
XML or tuples?
RDF and XML - which one do I choose?
A:
XML is a very powerful language for describing heiarchical things.
Defining what goes into a document - known target intereactions XML
does this very well.
Encoding all the semantics in human-readable terms: it doesn't mean anything to a machine - what that element is and how it relates to any other element - doesn't mean anything at all to machines.
If you or I read it we might think that name was related to the person; a date was a refinement. But structurally, and syntactically, they look the same. Unless preco-ordinated, understanding it is very hard - implicit, not explicit semnatics.
Charles: You can add a plain text documentation about what it means.
Danbri - there a difference in the perspectives of RDF and XML:
XML is about documents; RDF is about the world.
The number of elements of certain types versus people have names
etc etc; not the documents that describe them.
ericM - RDF is a relational model; xml is heirarchical
grant - the problem with the XML hierarchy is that it mixes 2 things
part/whole (e.g. names)
genus/species (e.g. dates)
ericM - yes, so have to precoordinate
danbri -
wonders if people have experienc of mixing dtds, xml schemas
[shows slide]
shows a diagram showing how can have modularity and fine-grained mixing of vocabularies, which is harder in XML. RDF can devolve creation of vocabularies to small groups.
ericM - conversations with XML people suggest this is very difficult problem, for which XML was not designed.
danbri - flip side is that RDF is unpredictable
ericM - and syntactic validation is very difficult.
Charles - you can have syntactically valid XML with a schema that also happens to be RDF - forward looking
[This discussion was supported by Tom Baker's slides -CMN]
The status of translations of DC - the idea is to distribute the translations so they could be harvestsed in RDF; discussion of policy issues
Haven't nailed doen what RDF code for this looks like; versioning; status.
The bit relevant for translation is the label, date, definition, comments - the stuff in english
EricM's proposal
<rdf:description rdf:about="URI">
<rdfs:label xml;lang:>label here</a>
<rdfs:comment xml;lang="">desc here</>
</rdf:description>
harry - missing provenance information
tom - does say who has done the translation or when exactly what was translated, e.g. when a standard is updated
Charles - can't you do this in the uri?
Tom - in principle yes:
<translates rdf:resource="http://dc/2002/10/1....">
eericM - this version is designed to be very simple, non-disruptive
harry - minimises confusion - no resources here.
danbri - could get arbitrarily complex - would like to get a description of a (student) project for building a harvester thing.
---
Roland's proposal
much more complex: provenance, bags (see Tom's slides - scribe couldn't copy it all down fast enough)
--
Tom wants to push the translatiosn out to the translators, and make them available for harvesting.
The simple version means that we have to tolerate that the translation of 1.1.x to a different language might differ slighlt from the English language version.
Tom - a simple java applet which calls up existing RDF schema for DC - you see new fields and then the translator can add to it
harry - serverside app to save on own site - thinks that mitzuhara?? has got something similar already.
EricM - need to make sure that harvestors can find it.
Simon - the only version that had authority for DC is the English language verion? (Harry said this)
Tom - evaluation in different languages is difficult (if don't know the language) but if could, then could endorse it in some sense.
danbri - looking for something easy to deploy soon.
[ shows a documen using seealsos to point to the translations]
any good, Tom?
Tom - a central translations page? - would need volunteers to maintain links to diff erentlanguages. This is like stage 1.5....
danbri: this could serve this role