Print

Print


Orion Buckminster Montoya wrote:
>It's always a possibility to make my XPointers more generic -- maybe
>rather than pointing to a specific phrase I could just point to the
>sentence or paragraph that contains it.  But I'd like for this to be
>as granular as possible, with the most succinct, and preferably
>somewhat automatable, markup possible.  It's not out of the question
>for a single person (me) to look at the revised text before it becomes
>the master document, and make whatever markup changes are necessary to
>ensure consistency for the XPointers -- but that seems like a bad
>thing to build in by design, and it would limit the eventual general
>usefulness of the system.  I'm honestly more inclined to generate an
>ugly, inhuman, very un-TEI UUID for all the node items' ids, rather
>than require human intervention on this level.

You might perhaps be interested in the strategy I used in my APE project 
("http://www.philo.de/ape/").  I faced a similar problem and my approach to 
solve it was to use URNs which are resolved in multiple steps into an 
actual link.  The idea is roughly as follows:  Firstly, one defines a 
canonical reference system (i.e. the URNs) for the primary source.  A 
catalogue "knows" the mappings between the URNs and the URLs of the source 
document (you may use XPointers here).  The annotations are using only the 
URNs.  In order to resolve a URN, the software looks up the URLs in the 
catalogue.  So if anything changes, only the catalogue needs to be 
updated.  (Actually, the reference resolution process in APE is more 
sophisticated, because it queries different sorts of catalogues to allow to 
display alternative views of a document --say, a transcription and a 
facsimile-- or to link one annotation to several text passages, etc.)

The description of the program on the Web site is a bit short, but an 
article, which explains the linking idea in detail, is to be published 
soon.  If you are interested, drop me a line and I sent you a preprint.

Dieter Köhler