|
XML-based text & graphics integration
|
 |
 |  | Find |
|
Abstract
|
Based on the Microsoft Office 2000 package and aligned with the current
SGML editing system, the 'Office Solution' presented here shows the next evolutionary
step in the application of SGML/XML and Web technologies. The development
of the solution focused on intelligent, interactive graphics in combination
with a proprietary navigation system and an effective model for linking text
and graphics. The system is integrated into a higher level information and
management system with defined access rights. ActiveX Control cannot be implemented
at the present time due to security reasons.
|
Contents
|
Background
Figure 1
. SGML editing system
In October 1998, an SGML-based editing system was introduced in the
Information and Communication Networks Group at Siemens AG for the purpose
of making the internal guidelines, work instructions and circulars available
in the Intranet in HTML format.
The system concept and prototype were presented at SGML/XML Europe '98
in Paris [MKHParis98].
The following aspects of the editing system used have remained open
because of future extensions to HTML and XML functionality – including
graphics options – by the W3C.
- Integration of active, intelligent graphics
- Introduction of an effective link model
- Migration to XML
Previous
Table of Contents
New XML based system
During XML structural design, it was demonstrated that DTDs can be greatly
simplified. Complicated link constructs originating in proprietary SGML data
entry tools were completely redesigned. The premise on which the new structural
design was based was that it uses few elements, attributes and content modeling
in a simple main structure including graphical components. The structure should
be applicable to many types of documents and it should be easy to edit the
documents. More complex structures should apply these definitions to form
the basic component.
The evaluation of different graphic formats and the task of finding
a solution based on Microsoft Office 2000 yielded the following conclusions:
- Microsoft Office 2000 should be used as a text and graphics editor.
- Graphics shoud be integrated as an XML application – i.e., as VML. Microsoft Internet Explorer 5 can display VML without any plug-ins.
- Links should be extracted from text and graphical documents and edited in a separate document.
These aspects are considered in the XML system design. By structuring
and correcting the Office 2000 output with special transformation scripts
(Balise), valid XML can be generated from text and well-formed XML from graphics.
Figure 2
. XML text/graphics system
Text component
Since Word 2000 is used as a typographical editor and not as a structure
editor, text must be transformed into parsed, valid XML. This approach guarantees
the automated editing functionality within a predefined structure.
Figure 3
. XML text component
Text must be written in Word 2000 in compliance with specific restrictions
– i.e., pre-defined paragraph and character formats must be used - and
text must be saved as RTF (Rich Text Format).
RTF documents are converted with a Balise transformation script using
a typographically oriented RTF-DTD to XML instances conformant with the desired
main structure of a given user DTD.
The resulting XML text documents do not contain any executable links.
However, all places that may be related to links (potential anchors) are marked
in the XML documents including nessessary information for the link model.
In addition, for semantic description of the XML document RDF metadata
(Resource Description Framework, Dublin Core) must be generated in XML syntax
(special metadata editor).
Graphics component
VML and SVG are both XML applications. The graphics
described with VML or SVG lose their static property (pixels) and are used
for displaying information intelligently as is also possible with CGM.
A prototype system was developed in summer 1999 including CGM and VML
to provide functions ranging from the generation of the participating components
to processability and display options. The goal was to find a suitable format
for integration in the XML system.
Graphical subcomponents can be selectively identified within an overall
graphic and linked as units of information to one another and to text components.
The text components can be XML elements from a text structured with a DTD
or caption text taken from a graphic.
Based on a concrete example of an XML document, the prototype demonstrated
the integrated use of VML and CGM:
- Scaling and navigation in VML and CGM components
- Separate manipulation of graphics and graphics text
- Linking of text components and graphics components
Microsoft Internet Explorer 5 was used for the display. All conversions
were performed using the Balise transformation tool.
The ISO-Draw tool was used for CGM graphics; PowerPoint 2000 and VISIO
2000 were used for VML graphics.
The results of using VML and CGM were as follows:
- CGM can be shown using plug-ins (IE5 plus Netscape) with predefined scaling and viewing options.
- VML can be displayed directly in the Web browser without a plug-in.
- Thanks to a mathematical model, VML and CGM graphics are scalable.
- VML can be shown quickly in the browser.
- Usable VML can be generated with the aid of standard tools (Office 2000, VISIO 2000) and proprietary manipulations.
- Additional graphic formats, including DRW, WMF, EMZ and CGM can be converted to VML using suitable filters.
- SVG can presumably be used as an alternative without a great deal of additional expenditure.
- SVG and VML as XML applications can be integrated in an XML system more homogeneously than CGM.
The findings of the prototype and the premises had the effect of focusing
document contents on the VML graphics format and on Microsoft Internet Explorer
5.
Figure 4
. XML graphics component
Graphics referenced in a text document are compiled as a set of slides
in PowerPoint. Each slide can be referenced as a graphic. The slides may contain
different kinds of information: vector, pixel and text.
When saving slides in PowerPoint 2000 (Save as Web Page...) an HTML
file is generated for each slide. In addition to the VML data the HTML files
also contain text and image data. A set of slides generated in this manner
is suitable for simple slide presentations using Microsoft Internet Explorer
5 but not for further editing and integration into the XML system.
Using a special balise transformation the HTML files were transformed
into well-formed XML.
Graphics are to be regarded on the one hand as supplementary modules
for structured XML text documents but, on the other hand, they can be used
as individual graphic files. Consequently, as with text, the use of metadata
(RDF) is imperative.
Link component
The use of an effective link model appears to be particularly important
when integrating text and graphics.
The quality of linked information objects depends for the most part
on how valid the links are – no sources without associated targets.
A common situation in the production of structured documents is the
separation of content and typography. As a result, the same content may be
represented in different ways. Breaking down documents into its constituent
parts of contents, typography and link information grants authors and designers
even greater flexibility.
Different types of links may occur in the link model used. Along with
the simple unidirectional links, there are also bidirectional and multidirectional
links. A link target can be a whole document or a particular location in a
document. Specific descriptions for sources and targets of links in documents
– so-called potential anchors – are inserted
in the document contents during the production of text and graphics. In Word
2000 and PowerPoint 2000 the options for labeling links are used in compliance
with relevant conventions to label potential anchors. A special element from
the user DTD is generated for each potential anchor in the text instance,
while a special attribute is generated in the graphic modules (VML) written
in XML. Whether these labeled positions are finally used for links is determined
in a subsequent processing step.
The fact that the author of a text and the graphics designer may well
be different individuals should be regarded. Moreover, the times at which
the individual components are completed may also differ. The same applies
in particular for pure text information systems (joint editing).
It is recommended to separate link information from text and graphics.
Links can therefore only be finally implemented once the text, graphics and
link definition are available.
Figure 5
. Editor - Link Definition Document
The link behaviour used is written using XML notation – the so-called
LDD (Link Definition Document) – and is separately administered. The
contents of an LDD are oriented according to a special DTD containing elements
that define potential anchors. Secondly, the DTD contains elements for potential
arcs representing the possible link connections. The entire link
information in the XML system is written "out of line", while incorporating
the principles of XLink/XPointer.
The actual document content (XML/VML) contains only locations for potential
anchors. These locations are automatically extracted to the latest LDD during
processing.
This link model allows individual, central administration of links (Link
Editor). Before generating the Web data, all potential anchors can be extracted
to the LDD from the documents involved in the generation process. Potential
arcs can now be edited more easily with regard to all potential anchors.
It is, of course, also possible to first construct a system of potential
arcs regardless of the availability of the actual associated text and graphics.
This method of link designing can also be a basis for efficient processing
and can be applied accordingly when assigning potential anchors.
Documents with identical content structures can share a link model (for
example, subdocuments or language variants).
Visualization on the Web
Finally, when all nessessary information is represented in XML, it can
be basically used for generating multiple results. The main goal of the system
however is to produce effective Web data enabling the use of new graphic and
linking mechanisms.
Figure 6
. Generation process
An information and navigation system based on HTML is generated using
a special balise transformation.
Results of the Web data generation:
- Corporate design of Web pages
- Detailed linking between text and graphics
- Valid link systems
- Definable scaling options for graphics
- Navigation in graphics
- Provision of typography in the form of referenced CSS
- Generated pages: index, glossary, table of contents, table of figures, etc.
Figure 7
. Navigation
The Web data is based at present on HTML. Should the browsers and the
XML standards prove to be sufficiently powerful, direct XML data solutions
can be selected. Note, however, that the major benefit of XML is processing
data. The support of different browsers makes generation variants necessary,
which adversely affects the functionality.
Conclusion
Figure 8
. System
The decision to use XML is to be regarded as an evolutionary approach.
The advantages of using XML-based graphics provide new aspects on the designing
of information systems. The solution presented here must be adapted to the
existing system environment. In this regard, questions of access security
and the tasks of the document management system must be addressed. A wide
application of the RDF model can be of great use for the future.
Bibliography