Visualizing Overlapping Hierarchies in Textual Markup

Visualizing Overlapping Hierarchies in Textual Markup

Patrick
Durusau

Society of Biblical Literature
pdurusau@emory.edu

Matthew
Brook
O'Donnell

OpenText.org
mtrout@nycap.rr.com

2002

University of Tübingen

Tübingen

ALLC/ACH 2002

editor

Harald
Fuchs

encoder

Sara
A.
Schmidt

The impact of SGML and XML on computing in the humanities is wellknown. Markup
languages have been used on such diverse texts as Nordic runes (Driscoll Driscoll, M. J., Encoding Old Norse-Icelandic primary
sources using TEI-conformant SGML/XML: A handbook ()), to the Oxford English Dictionary(New OED and Text Research UW Centre for the New OED and Text Research (ptr
target="http://db.uwaterloo.ca/OED/" />). Substantial experience has
been gained in the application of markup languages to diverse texts with
corresponding gains in the usefulness of such texts for computing humanists.
The default use of SGML for the representation of only one hierarchy, and the
explicit prohibition of overlapping elements in XML, has posed considerable
limitations to humanists attempting to record their observations of multiple
overlapping hierarchies in textual materials. Several approaches have addressed
these limitations, ranging from milestone elements(TEI), stand-off markup to
record hierarchies (Thomspon 1997Thompson, H.S., and McKelvie, D,
Hyperlink semantics for standoff markup of read-only
documents, Language Technology Group, HCRC, University of
Edinburgh()), fragmentation of a base text (TEI), to the development of
non-SGML/XML markup languages which will require special processing software
TexMECS: New meta-language for recording multiple hierarchies (Sperberg-McQueen
and Huitfeldt 2001Huitfeldt, C., and Sperberg-McQueen, C.M., TexMECS: An experimental markup meta-language for complex
documents ()).All of these approaches suffer from a variety of limitations and
problems. The most notable problem is a lack of commonly available software to
create and make the overlapping hierarchies available to scholarly
investigation.
Durusau and O'Donnell (2001Durusau, P., and O'Donnell, M.B., Implementing Concurrent Markup in XML, Extreme Markup
2001. Paper submitted to Markup Languages: Theory and
Practice. Presentation materials andworking examples available
at .) present a
method for recording overlapping hierarchies using XML standards, particularly
XPath and XSLT. The essential insight is that the key piece of information in
encoding a complex text is membership of a particular piece of text (PCDATA in
SGML/XML terminology) in a given hierarchy, and not its surface representation
through markup. Whether this membership information is represented in a
conventional SGML/XML "tree" or by some other means, it is the information that
is of interest to the computing humanist. This technique relies upon the use of
XPath syntax to record the hierarchy information for atoms of text in a larger
work. For example:
<w id="w1" sn:clauses="/clauses/clause[1][@id='c1']/s[1]/w[1]"
tx:text="/text/para[1][@id='p1']/w[1]"
pg:pages="/pages/page[1][@id='p1']/line[1][@id='l1']/w[1]"
>This</w>
Durusau and O'Donnell demonstrated the use XSLT to construct and then query
overlapping hierarchies of XML markup on a base text.The theory and practice of
querying of XML documents has been under rapid development at the World Wide Web
Consortium (XQuerySee ). Despite that pace of development, there has been no systematic
investigation of querying texts across multiple hierarchies. However, most of
the techniques currently under development recognize the importance of XPath
syntax on the effective formulation of an XML query language.
To develop effective querying of XML documents across multiple hierarchies,
Durusau and O'Donnell are developing techniques to visually present overlapping
hierarchies to allow "discovery" of fruitful lines of investigation for
particular texts. XSLT andSVG are used for this purpose. Overlapping hierarchies
may well exist for a text that are of no interest to one researcher that are the
focus of research for another. The ability to effectively choose which
hierarchies are of interest will be crucial to effective use of overlapping
hierarchies recorded for a particular text.
In addition to allowing a separation of hierarchies of interest from those that
are not, the visual presentation of overlapping hierarchies is also the first
step towards addressing the need for intimate knowledge of the underlying markup
to formulate a query for the multiple hierarchies recorded using this technique.
Effective queries now require mastery of the various hierarchies and a working
knowledge of their relationships, one to the other, in order to obtain any
predictable results. With an entirely visual presentation of the overlapping
hierarchies, scholars will be able to choose the hierarchies by display and not
markup syntax, as being the subject(s) of further analysis. This paper
demonstrates the use of XSLT and SVG to produce visual representations of
documents with overlapping hierarchies and the linking of these visualizations
to XPath and XQuery statements. Multiple editions, annotations and analyses of
Book 1 of Milton's Paradise Lost form the test case for
these techniques. The more detailed the analysis the greater the need for
detailed knowledge of the syntax but that will be only after the scholar has
found information of interest and not a general exercise in mastering several
divergent markup regimes. This will be of particular importance with many texts
such as biblical texts that have complex syntax, grammatical, literary and
transmission hierarchies that overlap and combine in complex ways.