|
|
 |
Article:
 |
 |
To Tag or Not to Tag
|
| Subject: |
Can TEI/XML really address the problem? |
| Date: |
2004-05-29 00:46:02 |
| From: |
Sancho Neves-Graca |
|
Response to: Can TEI/XML really address the problem?
|
|
|
You suggest that XML might not be able to cope with complex documents because of its tree structure. The tree structure is not per se a limiting structural aspect of XML documents. What constrains the structure of an XML document is its schema, what guides the processing are path values and linking elements, and what does the processing are the parsers and transformators. The constraints you indicate are well within the practices of experienced XML developers. It might help you to think whether some mathematical structure with a known visualization can be described in XML, and whether mathematical images described in separate XML documents could be cross-referenced or merged. With the soup of XML technologies (do not reduce your view of XML just to the XML trees!), XML, XSL, XPath, XPointer, etc, many of these structures can be represented. It sure won't be NVS that will break the limits of the XML toolset. |
Previous Message
|
Up
|
Next Message
|
- Can TEI/XML really address the problem?
2004-05-29 01:42:22 alexbrn
[Reply]
Yes, I am suggesting that the tree structure of XML is a problem for complex documents, because complex documents aren't trees - see the Patrick Durusau JITTs paper for a fuller discussion of this.
And yes, schemas allow the specification of some constraints on XML structures, but there are no standard XML schema languages that address the problems of modeling overlapping hierarchies (SGML with CONCUR comes closest).
Of course XML can be coerced into modeling sets or overlapping structures using techniques like milestone markup, but then the impedance mismatch between the structure being modeled and the underlying tree structure means that you're going hard against the grain when using the other XML technologies (XSLT etc.).
However, what I was trying to get at was that a solution is to reduce the problem where the "text" isn't modelled as a single instance, but as a set of instances (i.e. one per edition). However the processing solutions for that set then move mostly out of the XML/standards world and become application specific.
Thinking about it some more, it's possible there'd need to be some sort of abstract instance representing the 'plan' of a superset of all editions of any play, which editions are then synchronised with (using XPath/XPointer maybe)...
- Alex Brown.
|
 |
Sponsored By:
|