Minutes of the TEI Overlap SIG Meeting in Sofia, 29 October 2005

Minutes of the TEI Overlap SIG Meeting in Sofia, 29 October 2005

In Attendance:

  • Syd Bauman
  • Tone Merete Bruvik
  • Arianna Ciula
  • Juan Garcés
  • Ohya, Kazushi
  • John Lavajnino
  • Dot Porter
  • Andreas Witt
  • John Young

This was the third meeting of the TEI Overlap SIG, first convened at the Third Annual TEI Member's Meeting in Nancy, France.

Discussion at the meeting was mainly held within the context of the forthcoming chapter on Multiple Hierarchies for the P5 Guidelines, authored by Andreas Witt. Andreas took the SIG through the main approaches for dealing with overlap, most of which are discussed, though briefly, in the Concurrent Hierarchies chapter for P4.
  • milestone elements: empty elements which mark the boundaries between elements in a non-nesting structure
  • fragmentation of an item: the division of what logically is a single element into two or more parts, each of which nests properly within its context
  • virtual joins: the recreation of a virtual element from fragments of text, possibly discontiguous or out of order (aka stand-off markup)

In addition, we discussed other approaches that may be adopted by TEI and included in future editions of the P5 CH chapter.

  1. Non-XML syntax:

    LMNL (Layered Markup and Annotation Language): "LMNL, pronounced "liminal", is an experimental approach to digital text encoding that supports, in SGML/XML terms, overlapping elements (ranges in LMNL) and structured attributes (annotations in LMNL)." ( http://www.lmnl.net/ )

    TexMECS, An experimental markup meta-language for complex documents ( http://decentius.aksis.uib.no/mlcd/2003/Papers/texmecs.html )

  2. HORSE/CLIX

    See the OL SIG Wiki at http://www.tei-c.org/wiki/index.php/SIG:Overlap#HORSE

    We spent quite a bit of time discussing the attractions and pitfalls of HORSE/CLIX in the context of TEI. Essentially, this approach calls for breaking hierarchies that overlap from regular start and end tags, with verifiable content (<p>…</p>) into empty elements that mark the beginning and end of the content; thus, they do not actually contain the content. The starting and ending elements are linked together through matching values for "sid" (start id) on the opening tag and "eid" (end id) on the last tag (<p sid="id4">…<p eid="id4">).

    A major problem with HORSE/CLIX is that it may break the content models assigned in TEI.

    For example, a poem consists of groups of lines and sentences. Sentences begin and end in the middle of lines, and some sentences contain several lines while some lines contain several sentences. In TEI, <lg> contains <l> and <l> may contain <s>, but <lg> may not contain <s>. In this case, we need to ensure that the HORSing/CLIXing does not affect the defined content models for <lg>.

    GOOD (<lg> may contain <l> but may not contain <s>)

    <lg>
    <l><s sid="id1"/>…</l>
    <l>…<s eid="id1"/><s sid="id2"/>…</l>
    <l>……</l>
    <l>…<s eid="id2"/><s sid="id3"/>…</l>
    etc.
    </lg>

    BAD (<lg> may not contain <s>)

    <lg>
    <l sid="id1"/><s>…<l eid="id1"/>
    <l sid="id2"/>…</s><s>…<l eid="id2"/>
    <l sid="id3"/>….<l eid="id3"/>
    <l sid="id4"/>…<s></s>…<l eid="id4"/>
    etc.
    </lg>

    This approach places a great deal of responsibility on the editor, or requires a processor capable of automatically determining which elements can validly HORSE. In some cases, the valid HORSE may be the more complicated one; for example, an editor wants to mark every word in a poem when in one instance a word breaks between lines. <lg> may not contain <w>, so if we break <l> (the simpler choice), our TEI is invalid:

    <lg>
    <l sid="id1"/><w>…</w> <w>…</w> <w>…</w><l eid="id1"/>
    <l sid="id2"/><w>…</w> <w>…</w> <w>…<l eid="id2"/>
    <l sid="id3"/>…</w> <w>…</w> <w>…</w><l eid="id3"/>
    </lg>

  3. Existing special elements: <addSpan/> and <delSpan/> serve to mark the beginnings of instances of text that is likely to overlap with other structures. They are empty elements. The end of the Span is identified usually by an <anchor> with corresponding ID.
  4. We discussed the difference between milestone elements (which mark a point in the text) and empty elements that mark the beginning and end of spans of text (as opposed to regular opening and closing elements which act as physical boundaries to their content).

Work for the Future

  1. Draft review of chapter – Andreas will make the chapter available for review prior to inclusion in the Guidelines.
  2. Possible future cooperation with the Standoff Markup Workgroup
  3. Continue discussion and investigate new approaches, via Wiki and Listserv