IF (and that is a very big if) you want this level of analysis at item
level, you should consider going a bit further. Assuming your example
below multiplied by a few thousand, the tags as used quickly become less
useful in their current state. It would make sense if you are going to
that much effort, to differentiate the two geogname tags with for
example a role attribute along the lines of "sendplace" "receiveplace"
(please come up with better terms...) as these will make the elements of
some use to a searcher. With your current version, a search on geogname
just retrieves places sent from and to, whereas I assume searchers would
be interested in one or the other. I have found that if you mark too
many things with simple tags at item level, they very quickly become of
little use. Quite possibly the geogname element is being used elsewhere
as well, and these uses should have some sort of attribute set too.
While in the example you could do a context search (find persname in
origination) to find the writer of the letter, how do you search for the
recipient? It is probably as easy (unless you are working in a text
editor with no macro facilities [which is plain masochism]) to insert
element and attribute as it is element alone, so this option should not
involve you in any more work (you ought also to consider a normal
attribute for your unitdate).
Once you have done a few hundred thousand item level descriptions you
might (as I did) get rather less enthusiastic for item level tagging,
something I only now do with calendar lists (extremely detailed
catalogues of medieval charters using attributes to distinguish persname
as witness, notary etc., place of issue of document and so forth.
# Richard Higgins
# Durham University Library
# Archives & Special Collections
# Palace Green
# DH1 3RN
# E-Mail: [log in to unmask]
Terry Catapano wrote:
> At NYPL we are currently drafting local guidelines for encoding item
> level description. For what it's worth, right now our approach to
> letters looks something like this:
> <origination><persname>Ramone, Joey</persname>
> <unittitle><genreform>Letter</genreform> : <geogname>New York,
> N.Y.</geogname>, to <persname>Cheetah Chrome</persname>,
> <geogname>Cleveland, Ohio</geogname>, <unitdate certainty="circa">1976
> May 13</unitdate></unittitle>
> Of course. depending on the desired level of control/granularity we
> could employ the "NORMAL", "SOURCE", "ROLE", "ENCODINGANALOG", and
> "AUTHFILENUMBER" attributes on the one hand, or, on the other, not
> tag the contents of <unittitle> at all beyond <unitdate>.
> Hope this helps. Comments and questions are welcome.
> Terry Catapano