[VoCamp Discuss] Welcome, and Introductions

[Catherine Dolbear]

Hi Ben,
I really liked the very last line of your use case, "this triple/graph
of triples is valid in the context of this evidence [graph]" I think
it's part of a wider problem that I've also hit: how to describe the
context in which the link or triple is valid (not necessarily based, as
you need, on evidence of the statement's truth, but just based on
whether the two URIs can be considered the same for a particular use
case/purpose - I have pretty much an atheist's view on web identity: I
don't think there is an absolute truth).

An example: DBpedia's "England" is a "consituent country of the UK"
while Geonames's "England" is a "primary administrative entity" - these
have been linked by owl:sameAs. If you're just interested in visiting
England as a tourist, this link might be absolutely fine; but if you're
a government statistician collecting statistics about England and Wales,
say, you can't link to DBpedia because of its subsequent, link to the
(incorrect) Geonames. (officially, England doesn't exist as  a primary
administrative entity - in this context, it does matter.)

There may be some scope for using a quad rather than triple, where the
fourth element gives an indication of the context/evidence/provenance of
the statement.... but I don't know...

Cathy

-----Original Message-----
From: discuss-bounces at lists.vocamp.org
[mailto:discuss-bounces at lists.vocamp.org] On Behalf Of Benjamin O'Steen
Sent: 22 September 2008 12:45
To: discuss
Subject: [VoCamp Discuss] Welcome, and Introductions

Hi,

I'm Ben O'Steen, and work for Oxford Uni Library Services (OULS) dealing
with digital archiving, modelling of research output and the
preservation and presentation of said research. One of the main services
is the Research Archive (~ Institutional repository) at
http://ora.ouls.ox.ac.uk but I am dealing with most of the live archival
of digital material, such as archiving the Google Books digitisation
images, research datasets, etc.

Things that are on my mind that are relevant to VoCamp:

1) Very interested to see "Evidence" semantics appear in Ian Davis's
introduction [ -> "... looking for collaborators for an evidence-based
model of genealogical research (or general historical research)"] - I am
very, very interested to push this forward and I have a good use case
for it, which I'll outline at the end.

2) Vocabularies and policies to enable the translation of strings to
things, with regards to the metadata associated with research output
(e.g. "John Smith" -> (fict.)http://people.ox.ac.uk/jsmith2008
"Department of History" -> (fict.)http://departments.ox.ac.uk/history,
or dc:subject "Science" -> dc:subject
http://lcsh.info/sh85118553#concept

 -> So, persistance and retention of subject vocabularies, privacy (e.g.
medical services/animal testing faculties), and how best to translate
(institutional) organisational structures into things on the web are of
concern to me.

3) Data -> How best to represent or register the availability of 2d/3d+
tabular datasets, when the data size is in the Gb->Tb->Pb areas - i.e.
develop/find/steal a vocab for describing bulk datasets, like the NASA
FITS format or even CSV data.

4) More of a very current itch to scratch for me, but modelling a
phonetics (audio) database and presenting this dataset in a useful and
semantically sensible way.

5) Representing books/paginated content - e.g. is there a better way
than RDF:Seq? How best to capture the page ranges for title pages,
contents, indexes, plates, tables, appendices, etc. Again, looking to
use things, rather than strings here - a label of "Title Page" isn't
good enough.

6) Time vocabularies - (some overlap with #1 above) vocab/scheme/plan to
note the time span for which a given triple or set of triples are valid
for.

7) Tagging vocabularies - delicious/digg/reddit representation

See you all at camp (it feels american to write something like that!)
and I'll likely be installed at the Far from Madding Crowd pub from when
I get out of work on Tuesday night too :) 


Ben


Evidence-based historical assertions: use-case
----------------------------------------------

One project I am working on, is turning a historical record of medieval
library catalogues into an electronic set of resources.
(http://www.history.ox.ac.uk/sharpe/index.htm#catalogues)

>From initial discussions, it was very quickly apparent that this
information is best presented as a graph - a library/private home/etc
contains 'books', each of which contain 1 or more 'works', and a work in
a given book may have annotations/marginalia.

Unlike today, the recorded books and the compilations of works they held
were much more bespoke. And instead of having clean, clear catalogues of
which works were present where and when, each asserted record - [library
'A' held book 'B', containing works 'C' to 'F' during time X] - is known
due to a piece of evidence. The evidence is pretty eclectic, ranging
from the most definite and most rare (written record of sale or a lib
catalogue record) to the more interesting, such as book 'B' was found to
have the same form of binding that library 'A' used in the Xth century,
or even more loose, book 'B' must have been in library 'A', because the
handwritten shelfmark on the front page is in the handwriting of the
librarian who sometimes worked at library 'A' during time X.

This means that a numerical scale for strength of evidence is not
useful, but the ability/vocab/scheme to be able to say something like
"this triple/graph of triples is valid in the context of this evidence
[graph]"

_______________________________________________
discuss mailing list
discuss at lists.vocamp.org
http://lists.vocamp.org/listinfo.cgi/discuss-vocamp.org
.


This email is only intended for the person to whom it is addressed and may contain confidential information. If you have received this email in error, please notify the sender and delete this email which must not be copied, distributed or disclosed to any other person.

Unless stated otherwise, the contents of this email are personal to the writer and do not represent the official view of Ordnance Survey. Nor can any contract be formed on Ordnance Survey's behalf via email. We reserve the right to monitor emails and attachments without prior notice.

Thank you for your cooperation.

Ordnance Survey
Romsey Road
Southampton SO16 4GU
Tel: 08456 050505
http://www.ordnancesurvey.co.uk