Brief summary of research over last five years 
(words 460 max. 450, excluding pubs.)



The development of the Amsterdam Hypermedia model, Hardman's PhD thesis 1998,
allowed the expression of a document format that went
beyond existing temporal-based multimedia and link-based web
pages to a richer document structure allowing a combination of these.
The work was highly influential in the development of the W3C (World
Wide Web Consortium - the organization behind the World Wide Web)
recommendation SMIL (Synchronized Multimedia Integration Language).
In addition, the separation of concerns such
as timing and structuring led to the inclusion of temporal
information in other languages (in particular SVG and XHTML) within
W3C.

The creation of a document format alone is insufficient.  A means of
creating the format is required through an authoring system.
Theoretical work contributed to the development of CMIFed - a
hypermedia document editor. 

In conjunction with the work in the W3C SYMM (Synchronized Multimedia)
working group, it became clear that rich, declarative document
formats are useful for expressing information to an end-user.  This
allows choices of temporal-based information for showing processes and
link-based information for allowing user-based selection.  From an
authoring perspective, however, even with tool-based assistance, an
authoring environment based purely on the specification of the
(low-level) components of the document is insufficient.
Instead, a completely new approach was needed where the author is
elevated to a position of specifying the ingredients of a presentation
and the detailed specifications left entirely to the machine.

In addition to the difficulties of authoring any single presentation
for a particular platform, the number of forms of computers has
increased, with the corresponding desire to display a
web page on, for example, a mobile phone, a hand-held device or a
desk-top high-resolution screen.  This diversity of target platforms
means that some way of tailoring the information to the device is
needed.  Work has been carried out to automatically
generate presentations for different device characteristics.

In parallel with this work, the Semantic Web has been developing,
allowing the combination of existing Web document engineering techniques
with more knowledge-intensivee approaches.  The current state of the
research is that a number of projects are investigating aspects around
the core issues:

what types of metadata are required for multimedia (IEEE-MM
publication in press); 

how can results of database queries be expressed as meaningful
presentations (TI Topia project); 

how the user be included more explicitly in the process (NWO ToKeN2000
CHIME project);

what are good presentation generation approaches (NWO NASH project);

how can we capture knowledge on discourse structures and modalities and
incorporate this into the presentation generation process (NWO
ToKeN2000 I2RP project).

As current projects are beginning to show results, it is
becoming more clear what the exact problems are that need to be
tackled.  These problems are large and their solution requires a
complete re-examination of the document engineering approach. 


Key publications:


That Obscure Object of Desire: Multimedia Metadata on the Web
IEEE-MM to appear 2004