SIMILE minutes 23 october 2003

Attendees:

Mick Bass, David Karger, Rob Tansley, Andy Seaborne, Kevin Smathers, Mark
Butler

Regrets:

MacKenzie Smith, Eric Miller

mickBass: I want to take a poll of moving the plenary to the 17th and 18th
of December
David: That looks pretty good. I may have an hour of conflict here or there.
Mark: I'd have a preference of the 17th and 18th.
AndyS: That's okay.
Rob: Those dates work for me.
mickBass: I'll wait to hear from Eric, but keep the 16th - 18th open until
we can pick.
mickBass: I brainstormed a list of open actions from prior weeks.
One is to get a quick status on each of these items, the other is to see if
there are other actions.
Mark, you were looking to hook-up to Martin Doerr?
Mark: haven't done it yet.
mickBass: MacKenzie has forwarded a set of webservices we can take advantage
of, does it make sense to use them?
AndyS: I think we need to get the local corpus up and running, then refine
it. So the OCLC services come in at a later stage.
mickBass: David, this might be a candidate for a plug-in available via
HayStack
David: Thats the kind of thing Nick M has been working on.
mickBass: Perhaps we should forward MacKenzie's note to him?
Andy, are we all on the same page about what we really want to happen here?
AndyS: I thought MacKenzie / EM were referring to someone who had knowledge
of VRA and Artstor. It would be useful to talk to someone 
with VRA experience, just to walk through our decisions
I think both Mark and I are happy it can be done, but well enough for the
demo
mickBass: Maybe the followup here is do we have a sucinct list of the
modelling problems?
(Mark: Did anybody catch the name of David K's student who wrote the
ontology browser?)
mickBass: I'll let MacKenzie and Eric were are still looking forward to
review activities around the modelling choices
(Discussion on hiring / finances - omitted from public minutes)
mickBass: History system handoff - still at the same place
mickBass: David, Mark and Mick had a meeting to discuss possible work for
Steve Garland.
ACTION FOR MARK: Contact Steve Garland to discuss this
David: Steve is working on a help system for Haystack, accessing Haystack
over the web, and also building viewers for Artstor VRA, IMS and CIDOC data,
and looking at how to do delivery via a web client
mickBass: Questions and comments?
mickBass: MacKenzie has written back saying we have got all the course data
we are going to get, we need to find out
from Eric latest status on Getty and CIDOC corpus
One item I'd like to defer to on list discussion. Is there anything else we
want to talk about on the call?
kevins2: One problem with IMS is it doesn't really contain enough
information to create a canonical URI. They come in multiple forms, and I'm
getting annoyed with XSLT's ability to do string transformations, its no
where near as good as PERL
also I think we need to do lookups on a database, and doing calls out to
Joseki, rather than trying to do it in XSLT.
Although I have a fairly complete script for taking the XML and turning it
via to XSLT into RDF, but you get lots of duplicated people.
Creating canonical URIs for resources, but the problem is the entry in IMS
is a path name not a URL, so there is a standard translation and whats
in the file except sometimes its different. I know where the exceptions are,
but you need to test and not sure if you can do this in XSLT.
David: Nick M is working on this, I'd love to get him involved in this
discussion.
He's working on bibliography records.
kevins2: There is one other thing I want to discuss, the IMS schema using
RDF specifically asks for VCARD entities.
AndyS: I think VCARD would be acceptable. What we are talking about is how
to describe people by reference, so sometimes you can resolve that to
a URI, sometimes its just anonymous. 
kevins2: It's supposed to be a complete VCard, but in practice its a full
name field.
mickBass: One of the issues seem to be 1) where is the most appropriate
place to do this e.g. do it in XSLT, do it Perl, do it in RDF
kevins2: So when we get it into RDF we will have to get rid of duplicate
records
mickBass: we do have some time to market pressure here. The other is
contribution to the research questions here.
E.g. I've got all this XML data, how do I make it available in RDF. So that
would led to get it into the RDF domain.
AndyS: once its in RDF, you could use the rules in Jena to do some
processing. Generally we are trying to get the corpus established. 
marbut: 1. Kevin, if you could send descriptions of what you've done, I may
be able to make suggestions in XSLT domain
2. Big problem working with internationalization
(this is an issue with ArtStor corpus)
... with XSLT that is
3. One approach is using RDDL, and specifying XSLT stylesheet in the RDDL
doc
this may be more flexible because everything can happen at resolve time
rather than with perl which won't do this
4. N3 seems more usable than RDF/XML serialization
kevins: but you can't output N3 via XSLT

Received on Thursday, 23 October 2003 11:49:12 UTC