[open-literature] Fwd: http://www.opencorrespondence.org/

[Rufus Pollock]

Forwarding James' message to mailing list so we can discuss it here ...

---------- Forwarded message ----------
From: James Cummings
Date: 22 July 2010 12:39
Subject: http://www.opencorrespondence.org/
To: info at okfn.org

Hiya,

Some feedback about your opencorrespondence.org site:

a) Your 'Want to help out' link to
http://www.opencorrespondence.org/get-involved ends in a 404.
b) When looking at any individual letter your 'RDF version' is really
misleading as this is only RDF of the metadata concerning the letter.
I'm glad that there is open linked data (we're in the midst of doing
something similar for data.ox.ac.uk) but maybe call it 'RDF metadata'?
c) In rendering the letters as XHTML you use
<pre><code>...</code></pre> around what are commonly called openers
and closers in letters.  This is semantically really quite icky, and
as I'm sure you can appreciate those wanting to re-use the data would
prefer you tag it in something that makes more sense. (See d))
d) In reference to c) I don't know what underlying format you've
encoded the letters in before, one hopes, generating the XHTML pages,
but have you consider TEI P5 XML (http://www.tei-c.org/) as a
semantically rich format for the underlying storage of the letters? It
is a de facto international standard when it comes to marking up
historical texts. It has elements for all the textual distinctions you
would want to make, and more, and can easily be customised and
constrained to suit your needs and to give you a tighter schema. I'm
not saying you shouldn't use RDF as a linked data output, just that
the underlying storage format should be richer. I'd also encourage you
to release the underlying data for the letters in some structured,
preferably XML, format so those wanting to use the letters as a corpus
can do so.
e) Your letters provide the date 0000-00-00 when presumably the date
is unknown... wouldn't it be better to transform this to 'Unknown' or
similar?
f) http://www.opencorrespondence.org/letters/view/1 seems to be an error.
g) I'm happy to give you help and advice in any of the above. If you
have the letters as a whole in some XML/XHTML representation and want
an XSLT transformation to semantically-richer markup I would be happy
to attempt to write that for you.

Hope that helps,

-James

--
Dr James Cummings, Research Technologies Service
OUCS, University of Oxford