Thanks, Lou, for forwarding that mail from Richard Light!
Richard Light wrote:
> One challenge in playing this game, as I'm sure you are aware, is that
> the nature of such assertions within prose doesn't necessarily lend
> itself to neat hierarchical packaging. The <persEvent> example in
> 18.104.22.168 is nicely structured, but not all assertions that A
> was married
> to B on a certain date at a given place will be so well-behaved. To
> take two obvious cases: they could span two paragraphs; or they could
> contain extraneous text which would perforce have to fall,
> within the <persEvent> element.
This is true ... and this is why the persEvent element is allowed only
in the teiHeader - it's not applicable (sadly) to the task of adding
semantic markup directly to encodings of existing text.
> I see the tendency in P5 is towards rather specific person-oriented
> concepts (e.g. affiliation, floruit, persEvent), and would
> question the
> wisdom of this. A museum person or antiques dealer might
> come along to
> the account of the Morris marriage and say "hey, what about marking up
> those gifts they exchanged: they are interesting objects".
> (After all,
> the antique silver cup might now be in the Ashmolean: how do you make
> that link?) Someone with an interest in architecture might complain
> that the church wasn't identified as such in the markup, or a
> that you had no means of identifying it as a place within the town
> Oxford (or indeed within the county Oxfordshire!).
> I think the P5 markup scheme should enable a more agnostic approach to
> identifying real-world entities, the relationships between
> them, and the
> events in which they played a part. One obvious start would be to
> broaden out persEvent to just "event". This would allow you, for
> example, to mark up a subsequent paragraph which notes that
> the antique
> silver cup was stolen in 1863.
Personally I'm inclined to the view that "ontological" TEI markup should
start off based on a simple assertion-markup, perhaps using <graph> and
<node>, and that even such basic ontological concepts as "event" should
be defined in an external ontology, and referenced by URI. This is not
at all to prevent TEI customisations from defining a set of elements
which would be "semantic sugar" for the combinations of this generic TEI
element with different ontological categories, but I think a good
approach might be to establish TEI markup for this lowest level first,
which could be used without any TEI customisation (RDFa-style), and
subsequently to make TEI customisations to simplify encoding assertions
in this or that ontology. So we could then have a CIDOC CRM
customisation of TEI, for instance.
[log in to unmask]
New Zealand Electronic Text Centre