On 11 May 2005, at 01:38, Art - Arthit Suriyawongkul wrote:
> I'm trying to annotate a small piece of news in German (note, I cannot
> speak German) with TEI/TEI Lite.
> With the nature of the text type, it deals with lots of names,
> refering strings, and quotes.
> And I'm getting confused at some about usage of <q>, <rs>, <name>
> (and may be <persName>, <placeName>, ...) and how they can be linked
> to each other.
This is a tangled topic but there is an underlying logic in the way
these different elements are defined. The first thing to grasp is that
they are all to do with representations of names, not with
representations of entitites (such as persons, places etc.)
> For example, in an annotated text:
> <rs>Der <name type="person" key="RABI1" reg="Rabin,
> <name type="person" key="AMIJ1" reg="Amir, Yigal">Jigal
> hat am heutigen Montag morgen vor
> <rs>einem Gericht in <name key="TA1" type="place">Tel Aviv</name></rs>
> bei einem Haftpruefungstermin sein Gestaendnis
> wiederholt und bekraeftigt , den Mord an
> <rs key="RABI1">dem israelischen Ministerpraesidenten</rs>
> ohne Mittaeter veruebt zu haben .
> The first <rs> is "The Rabin-killer Yigal Amir",
> which contains two <name>s, "Rabin" and "Jigal Amir".
> And the second <rs> is "the Israeli prime minister".
My (also minimal) knowledge of German suggests you actually mean
> Can the second <rs> has a key "RABI1" that refer to the first <name>
> inside the first <rs>
> (the Israeli prime minister --> Rabin), please ?
Certainly. The point of the key attribute is to associate all the
different representations of some name with some normalised value,
which could be used as a key to search a database of entities (not
names of entities),
If you want your encoding to say that the name XXXX and the name YYYY
are actually both referring to the same person, you should do it with
the key attribute, as you are doing here.
> Or I have to use other attribute instead ?
> And if I like to use a more specific <persName> instead of generic
> in the case of "Rabin" it may look like this:
> <persName key="RAB1" reg="Rabin, Yitzhak">
That's also fine, though I think I would put the key attribute on the
outer <rs> for preference.
> but how to tell that only "Rabin" will be display, and not "Jitzhak" ?
That's up to your stylesheet. Nothing to do with the encoding. Ah, wait
a minute, I see what you mean. If the forename is not present in the
text you are encoding, then I don't think you should add it within the
persName element. If you want to know the regularised form of the name,
then that's what the REG attribute gives you (things are a little
different at P5. but that doesnt affect this issue). Otherwise, how can
you recover the actual form of the persName as given in your source?
> And if there's a followed sentence:
> <q who="AMIJ1" type="spoken"> Ich handelte allein , aber vielleicht
> mit Gott </q> , sagte <rs key="AMIJ1">der 27jährige Jurastudent</rs> .
> " '..direct speech..' , said the 27-year student. "
> can the 'who' attribute refer back to "AMIJ1" (<name>Jigal
> Amir</name>, above) as well ?
Yes. But note that the who attribute is defined in P4 as IDREF, so you
must actually define a <person> element with the ID value AMIJ1
somewhere (in the header, in a [participant description for example)
for this to work.
KEY -- supplies an identifier which allows you to associate things by
co-reference, without necessarily supplying any definition of the thing
they are co-referencing.
WHO - supplies an identifier which associates to something that must
exist in the current document. The identifier of that thing is provided
on its ID attribute,
You will say, why doesn't <q> have a KEY attribute then? Maybe it
should, but no-one has suggested it yet!
Hope this helps
> Thank you.
> :: Art / Arthit Suriyawongkul
> :: http://siit.net/members/art/ | http://bact.blogspot.com/
> Thai Wikipedia -- Free Encyclopedia
From the Macmini at Burnard Towers