From DTD generation to XML conversion:

1
From DTD generation to XML conversion

• Uwe MüllerHumboldt University, Berlin
• Electronic Publishing Groupu.mueller_at_cms.hu-berli
  n.de

2
Background

• Humboldt University 800 1.000 dissertations /
  year
• Germany duty to publish dissertations
• traditional methods
• publishing house
• microfiche
• 40 200 printed copies (depending on faculty
  regulations)
• Humboldt U. not mandatory to submit an ETD
• ¼ dissertations published electronically
• XML as central strategy

3
Why XML?
PDF

• Standardized format
• Long term preservation
• easily convertible to
• presentation formats (HTML, PDF)
• other XML structures
• qualified full text retrieval
• contains structural and contextual information
  in a machine readable format

digital signature
Office document
digital signature
XML
digital signature
HTML
4
XML Restrictions to deal with

• XML source does not contain layout information
• rather linear structure
• XML is not used as Authoring System
• authors use their 'own' systems
• Microsoft Word
• LaTeX
• Open Office / Star Office
• Framemaker
• Word Perfect

5
How to overjump the gap?

• get the authors where they are
• instructions and guidelines for authors
• usage of style files (e.g., dissertation-hu.dot)
• manuals, support hotline, regular courses
• different conversion processes
• SGML author (plug in for MS Word lt 97)
• Open Office / Star Office
• exploit genuine XML format
• MS Office 2003
• XML according to DiML DTD
• common pitfalls tables, pictures

6
(No Transcript)
7
Conversion Process Using Open Office
example.doc
front.html
chapter1.html
chapter2.html
Open Office
example.sxw (zip file) . . . . . . . .
chapter3.html
.gif
.jpg
example.html
content.xml
front.xml
example_stl.xml
chapter1.xml
chapter2.xml
chapter3.xml
example.xml
8
(No Transcript)
9
Principal Structure of a DiML document

• ltetdgt
• ltfrontgt..title...author...abstract...lt/frontgt
• ltbodygt
• ltchaptergt
• ltsectiongt
• ...
• lt/bodygt
• ltbackgt..bibliography...appendix...vita...lt/backgt
• lt/etdgt

10
From flat structure to Hierarchy

• only two types of styles in Word
• paragraph styles
• character styles
• e.g., in case of th first occurring Heading 1
  paragraph style the converter has to know
• Heading 1 is the beginning of a chapter
• Heading 1 implies a head element
• the element chapter can only occur in body
• lt/frontgt
• ltbodygt
• ltchaptergt
• lthead id"anyID"gtIntroductionlt/headgt

11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
One Core Multiple Views

• HTML generation (static or dynamic)
• performance problems with XSLT and huge documents
• solution division of XML sources into components
  (easier and fast to process)
• PDF Print on Demand (http//www.proprint-service
  .de)
• Current problems
• changing Office systems and versions
• ongoing implementations and adaptations necessary
• but might be restricted to XSL coding

18
Towards a universal DTD?

• DiML originally taken from an SGML DTD at
  Virginia Tech ("ETD"), http//edoc.hu-berlin.de/di
  ml
• already many elements (gt 100)
• combines elements of different description levels
• extended and adapted to local needs
• special requirements from several departments
  (e.g., literature / dramatics, humanities,
  geography, )
• necessity to include external DTDs (e.g.,
  CALS-Table, MathML, MusicML, )
• publication types other than theses and
  dissertations
• conference proceedings, electronic journals,
  other series,
• first approach extend DTD aiming at a universal
  'mega' DTD
• problems complexity, difficult maintenance
• other possibility create a completely new DTD
  for each purpose
• loss of interoperability

19
Modular DTD Approach

• idea individually adapted DTDs
• split up DTD into modules, such as
• text, structure, citation, dramatics
• handle external DTDs as modules as well, e.g.,
• MathML, MusicML, CALS-Table
• recombine a DTD out of user selected modules
• result
• a DTD with only the needed elements and modules
• individual reference and sample documents

20
Modular DTD Approach Benefits

• modules are easily maintainable
• distributed development
• version numbers for each module
• reusability
• define (several) styles for each module
• reference information for each module
• support different languages
• get a DTD that exactly fits your needs

21
DTDSys Principal Architecture

• modules small packages of elements belonging to
  each other
• stored in separate files in the DTDBase
• include metadata, e.g., descriptive information,
  version numbers, and dependences to other modules
• DTDSys generates DTD and reference files using
• XSL / XSLT
• Java
• Web Interfaces

22
Modules and Dependences

• text br, em, strong, sup, sub, u, tt, pre
• common p, head, caption, url, name, foreign
• structure chapter, section, subsection
• citation quotations and references
• documents page numbers, footnotes, endnotes,
• diml front, body, back, abstract

23
DTD Generation Process
XSL
dependences.html
xdiml.dtd
DTDBase
XSL
XSL
selection.xml
full-dtd.xml
module-text.xml
module-text.xml
XSL
module-text.xml
reference . . . . .
dtd-reference.xml

• including
• element info
• description
• dependences

p.php
JavaXSL
chapter.php
24
Outlook

• SCOPE Service Core for Open Publishing
  Environments
• development of Publication Components (authoring
  tools, conversion mechanisms, layout and style
  definitions)
• management system to maintain versions and
  dependences
• publication system
• workflow component
• Long Term Preservation activities
• Implementation of OAIS reference model
• Sun Center of Excellence

25
Thanks

• to Sabine Henneberger, Jakob Voß, Matthias Schulz
• Thank you!
• Questions?
• u.mueller_at_cms.hu-berlin.de
• http//edoc.hu-berlin.de/