From your description, I'm thinking the problems you describe would be a
perfect application for Topic Maps.
I don't know whether anyone's layered Topic Maps over TEI source yet; but
it's begging to be done. In this case I'd think you'd maintain all your
source files in TEI, with the global attributes providing no linking
semantics themselves, but only serving as anchors for links stored in the
topic map(s). The Topic Maps design is well-suited for this, and an
architectural argument may be made that whether you used Topic Maps tagging
for this or any other (e.g. an appropriate TEI subset with linking/indexing
semantics), it should be in a separate layer, not embedded in the TEI
There is a growing literature on Topic Maps, and a body of expertise is
beginning to coalesce after some years in the development/standardization
processes. See http://www.topicmaps.org and
At 09:11 PM 2/6/2003, you wrote:
>Please forgive the possible naivete -- and other standard disclaimers. In
>encoding a medieval exposition on another medieval text, I have two
>questions, which may be related.
>1. The author of the Encoded Text continually makes short quotations from
>the Referenced Text throughout his own writing. I am looking to encode
>this in such a way that the quotations in the Referenced Text may be linked
>to the appropriate place in the Referenced Text, which will be minimally
>encoded and provided as a reference point.
>Section 6.9 of the Guidelines seems to have what I want: reference
>systems. I might use <quote>, and encode the values I need (for example,
>Inf.01.070) into the n attribute.
>It is not clear to me, however, how I would go about logically connecting
>the n attributes in the Main Text and in the Referenced Text if the two are
>kept in different .xml files.
>2. We would like, eventually, to provide several different indices to the
>Encoded Text -- namely, an index of proper names, separated into people,
>places, works of literature and Other; and an index of major thematic terms
>discussed in the exposition. We would like those indices to be separate;
>and we would also like to link, for example, literary work names to the
>names of their authors, and quotations from them to both. Finally, some
>external quotes are brought into the Encoded Text without a mention of the
>work from which they are drawn -- in this case, my understanding is that we
>would still create an index entry for the work, placing it somewhere out of
>the way (where?), and then use level1-level4 to link the quotation up to
>its corresponding work title in the index.
>I looked at the documentation for <index> and at section 6.8.2 -- Index
>Entries. I did not find an indication there of how we would go about
>separating the encoded terms to be indexed into their respective
>indices. I am also not sure how we would cross-reference authors, their
>works and quotations as described above -- perhaps encode
><name><index>Person</index></name> and cross-reference that with an
>attribute in <quote>? (That would take care of connecting authors and
>quotations, but not works and quotations, or authors and works.)
>I suspect that the answers to my questions lie in global attributes, and
>confess that my understanding of those is rather shaky. I was unable to
>find information on global attributes, gathered in some one place --
>perhaps I am missing the obvious? Help on- or off-list would be appreciated.
>[log in to unmask]
"Thus I make my own use of the telegraph, without consulting
the directors, like the sparrows, which I perceive use it
extensively for a perch." -- Thoreau