Taxonomy question

Hello everybody. I would like to raise a question about the use of
the classification system provided by the <classDecl> element of the
<teiHeader>.

First, I must say that our group (http://www.deli.deusto.es/) is
using TEI with a somewhat heterodox purpose. We have a corpus
(hundreds) of little (from 1 to 10 pages) textual documents, but not
literature: they are of administrative kind (letters, contracts,
etc.), and a sufficient (however small) set of TEI tagging applies to
them, in the basic TEI spirit, in our opinion.

So, we have a complex taxonomy (+- 300 final categories in three
levels), and we asign exactly one final category to each document.

The question is: how to asign the category to each <TEI.2> separate
document, without enumerating the whole taxonomy in the <teiHeader>?

We are not experts at all in TEI, but reading the Guidelines-P4, we
could see some solutions, but we don't know if they follow the TEI
spirit.

Solution 1. If we put all the documents in a big corpus, we could
have the whole taxonomy in the <teiHeader> of the corpus, and then a
<catRef> in the <teiHeader> of each document.

OK, we plan to do that soon, but anyway *we also want* each document
in a separate TEI file, with his <catRef> identified.

Solution 2. Use <classCode>:

Re: Taxonomy question

I asked a similar question around Feb 18, 2003. Francois Lachance had a
valuable response in which he suggested something like your solution #2:

       <classDecl>
         <taxonomy id="general">
           <bibl>
             <title>General Taxonomy</title>
             <ref>http://www.foo.org/taxonomies/general/general.xml</ref>
           </bibl>
         </taxonomy>
       </classDecl>..

I had been thinking about using xptr instead of ref:

       <classDecl>
         <taxonomy id="general">
           <bibl>
             <title>General Taxonomy</title>
             <xptr url="http://www.foo.org/taxonomies/general/general.xml" />
           </bibl>
         </taxonomy>
       </classDecl>

I noted that adding a "url" attribute to xptr requires a statement in the
internal subset, but someone else pointed out that the Guidelines do
suggest that we do this.

j

numbers in prose

Humbul Call for Online Scholarly Humanities Resources (fwd)

Sometime ago staff at Humbul put some effort into cataloguing projects
using TEI. We are currently circulating a request to be notified of
online scholarly humanities resources (appended) and would be interested
to receive submissions from the TEI user community. Please feel free to
forward the call to colleagues.

Best wishes,

Michael Fraser
Head of Humbul
University of Oxford

---------- Forwarded message ----------
Date: Mon, 24 Mar 2003 21:10:16 +0000
From: Michael Fraser <mike.fraser <at> COMPUTING-SERVICES.OXFORD.AC.UK>

Call for Online Scholarly Humanities Resources

The Humbul Humanities Hub (http://www.humbul.ac.uk/) invites members of the
humanities research and teaching community to share information about
scholarly Web sites via its "Suggest a Resource" page at
http://www.humbul.ac.uk/submit/

Humbul catalogues online resources in the humanities, a remit that
includes History, Archaeology, Classics, Philosophy, Religion, English,
Modern Languages, Linguistics, and cognate subjects. As part of the
Resource Discovery Network (RDN), Humbul is a free service supporting
researchers, lecturers and students in higher and further education.
Please note that arts and creative industries subjects will be supported
by the Artifact hub (http://www.artifact.ac.uk/).

Scholarly resources suggested by colleagues assist our own resource
discovery activities and benefit humanities academics and students seeking
useful online resources.  Each suggested site which meets Humbul's
collection development policy, http://www.humbul.ac.uk/about/colldev.html,
is fully described by a subject specialist cataloguer and the resulting
metadata record is made available for searching and browsing.

In addition to building its catalogue, Humbul develops tools to make access
to its catalogue easier. The My Humbul suite of personalisation tools
(http://www.humbul.ac.uk/help/myhumbul.html) includes an email alerting
service and enables the reuse of Humbul's records you select within your
own web pages.

The Humbul Humanities Hub is a service of the Resource Discovery Network
funded by the Joint Information Systems Committee and the Arts and
Humanities Research Board, and is hosted by the University of Oxford.

Humbul Humanities Hub, University of Oxford, 13 Banbury Road, Oxford OX2
6NN. Tel: 01865 283 343. Fax: 01865 273 275. Email: info <at> humbul.ac.uk.

Re: numbers in prose

> With the Digital Library of Dutch Letters, Leyden, the
> Netherlands, we digitize paper-editions of older Dutch literary
> text. Sometimes those texts have prose with a kind of line
> numbering in the margin.

How you encode these depends a lot on how you think of them, and a
bit on what you want to do with them. If I am imagining what you have
correctly, I think of the line numbers in the margin are a sort of
referencing system, much like page numbers. In which case, one good
method would be to encode them as <fw>. The rend= attribute of <fw>
would allow you to specify they were in the margin:

  <fw type="lineNum" rend="place(left) pre([) post(])">17</fw>

> For us, using TEI-Lite, these numbers are hard to handle,

Oops. TEI Lite does not have <fw>. To me this is a good argument that
you should not be using Lite. There are probably some less desirable
solutions available in Lite, though.

E.g., one might think of the line numbers as sort of special-case
labels, and use

   <label rend="place(left) pre([) post(])">17</label>

Or use the n= attribute of each <lb>, or at least each <lb> which
actually had an associated line number in the margin:

   <lb n="17"/>      Or, I think less desirable:    <lb n="[17]"/>

(Note that you need to decide, and describe very clearly in the
<teiHeader>, whether the value of n= on <lb> refers to the preceding
or following lines sequence number. I.e.

| [1]  ... rights language used by those in power against those on
| [2]  the bottom should be treated with suspicion, while rights
| [3]  language used by those on the bottom against those in power
| [4]  should be treated with respect

could easily be encoded as either

  <p>
   <lb n="1"/>&hellip; rights language used by those in power against those on
   <lb n="2"/>the bottom should be treated with suspicion, while rights
   <lb n="3"/>language used by those on the bottom against those in power
   <lb n="4"/>should be treated with respect<sic corr="."></sic>
  </p>

or

  <p>
   &hellip; rights language used by those in power against those on<lb n="1"/>
   the bottom should be treated with suspicion, while rights<lb n="2"/>
   language used by those on the bottom against those in power<lb n="3"/>
   should be treated with respect<sic corr="."></sic><lb n="4"/>
  </p>

(Quote taken from end of the article that started at
http://www.arstechnica.com/wankerdesk/01q3/ip-ethics/ip-ethics-1.html.)

I think most folks would naturally go with the first example
(following), but I have no data to back up that assertion.

Note that the same difficulty applies with the use of <fw> or
<label>, but in those cases we would likely end up encoding what was
on the source page where it occurred on the source page.

> ... also in respect to a more or less functional and acceptable
> HTML-representation.

I will leave this to those who know more about HTML (a language I
know very little about).

> My question is: will TEI provide a simple solution for this kind of
> references?

Solution? Yes. Simple? You be the judge. 

BTW, as you consider how to encode this stuff, you may very well want
to think about what you do when there's an error.

   <fw type="lineNum" rend="place(right) pre([) post(])">
     <sic corr="182">128</sic>
   </fw>

(There's lots more to be said on the issue, to be sure, but I just
wanted to pique your interest. Gotta go ...)

Re: numbers in prose

On Fri, 2003-03-28 at 12:41, René van Stipriaan wrote:

> With the Digital Library of Dutch Letters, Leyden, the Netherlands, we
> digitize paper-editions of older Dutch literary text. Sometimes those
> texts have prose with a kind of line numbering in the margin. For us,
> using TEI-Lite, these numbers are hard to handle, also in respect to a
> more or less functional and acceptable HTML-representation.

Firstly, don't assume you have to use TEI Lite! the rest of
the TEI is just a few clicks away in the pizza chef.

Secondly, have you considered marking up line of text
in an element (eg <ab>) as if it was  a line of poetry?

Thirdly, how about marking the line endings with <lb n="345"/>?
that would transmute into <br> [345] in HTML rather easily.

I found Syd's suggestions a bit hard for my tiny brain....
--
Sebastian Rahtz      OUCS Information Manager
13 Banbury Road, Oxford OX2 6NN. Phone +44 1865 283431

Re: numbers in prose

On Friday March 28 Sebastian Rahtz wrote:

> Firstly, don't assume you have to use TEI Lite! the rest of
> the TEI is just a few clicks away in the pizza chef.

Too true! I wonder if the time hasn't come to "re-brand" or "re-position"
TEI-Lite to alter the image people have that this is the only TEI offering
that doesn't entail getting bogged down in arcane technicalities?

After all the "LITE"-ness predates the Pizza Chef, and since his baking is
now so reliable and his ordering system so straightforward, it's debatable
whether TEI Lite should still be marketed as the easy option.

I even wonder whether TEI Lite couldn't be removed as a separate
deliverable, and replaced by an option button on the Chef's menu that would
produce TEI-Lite with just two clicks. That way, people who weren't yet
ready to go beyond the Lite menu could still get their normal mix, but they
would need to go into the Bakery, realise it was now quite a warm and
friendly place and so be inclined to broader the scope of their orders on a
later visit. It would also make it plain that there is nothing special
(indeed not even anything especially "lite") about TEI-Lite, since it's just
one customization among many other possible ones, which are all equally
there for the asking.

Michael Beddow

Re: numbers in prose

> I even wonder whether TEI Lite couldn't be removed as a separate
> deliverable, and replaced by an option button on the Chef's menu
> that would produce TEI-Lite with just two clicks. That way,
> people who weren't yet ready to go beyond the Lite menu could
> still get their normal mix, but they would need to go into the
> Bakery, realise it was now quite a warm and friendly place and so
> be inclined to broader the scope of their orders on a later
> visit. It would also make it plain that there is nothing special
> (indeed not even anything especially "lite") about TEI-Lite,
> since it's just one customization among many other possible ones,
> which are all equally there for the asking.

I have been recommending essentially the same thing for years. (In my
version there's no special button for Lite, just that following the
baking process accepting the default value for everything would lead
to an output DTD that was TEI-Lite. A bit difficult to do -- remember
that TEI Lite does add several phrase-level elements (<gi>, <eg>,
<ident>, <code>, and <kw>), so the user would have to be walked
through the creation and use of extension files.)

Re: numbers in prose

> I have been recommending essentially the same thing for years. (In my
> version there's no special button for Lite, just that following the
> baking process accepting the default value for everything would lead
> to an output DTD that was TEI-Lite. A bit difficult to do -- remember
> that TEI Lite does add several phrase-level elements (<gi>, <eg>,
> <ident>, <code>, and <kw>), so the user would have to be walked
> through the creation and use of extension files.)

I'm probably missing something, but my notion was that the proposed "Lite"
button would transparently feed in the required extension files to generate
Lite. Isn't that practicable?

Michael Beddow

Re: numbers in prose

> ... my notion was that the proposed "Lite" button would
> transparently feed in the required extension files to generate
> Lite. Isn't that practicable?

Yes, I presume so. I am not a web-forms guru, but I can see no reason
why clicking the "Lite" button couldn't do that. It just wasn't part
of my proposal, simply because I hadn't thought of it.

But now that I do think of it, I'm not sure I don't still like my
proposal better -- to force the user to at least acknowledge the
existence of a pizza-baking process that can have extensions. Will
have to ponder it some more.