Description of the proposed research
(767 words, max. 800 words on max. 2 pages.) 

Title:
Effectively Conveying Information through Meaningful
Presentations

----

The Web has initiated a wave of research activities, many of which are
centered around finding relevant information on the internet.  Finding
relevant information alone, however, is insufficient to satisfy our
needs.  The retrieved information should be organised and visualized
in a manner suitable for effective human consumption. This problem is
illustrated by current generation search engines, which deploy
advanced techniques for finding relevant Web pages, but do little more
than presenting the results as a ranked list of links.

Contrast this with traditional media, such as newspapers and scientific
articles, where collecting the raw material is only the initial stage of
the complete information dissemination process.  The majority of the
work goes into structuring the information, making explicit
relationships to background information, highlighting important
information and presenting the results in an effective way.

While computers have become <i>de facto</i> information processing
machines, they currently provide little intelligent support for such
tasks.  <b>The challenge for computer science research is
to transform the computer from a passive rendering device to an active
participant in the process of conveying information effectively to the
end-user.</b>

The process of creating effective presentations has traditionally been
carried out by highly skilled people.  They are trained to balance
many different, potentailly contradictory,  interests (e.g. the use of
company colours vs. legibility). 
Computers are not yet able to generate presentations because
 most of the relevant theories that
analyse the way humans perform such tasks cannot be directly
translated into the formal and rigid computational models used to steer
computers.
The goal of this research is, to the extent that it improves automated presentation
generation, to capture knowledge from appropriate design professions,
to distill it and to incorporate it into a computational
environment.

In addition, the overall 
research goal requires the re-examination of the traditional document
engineering paradigm: that content,
document structure and graphic design can be determined independently
of one another.  The field of document engineering was originally
developed for predominantly text-based information.  Retaining the independence
of the three key aspects of a presentation does not allow the required
highly flexible construction of presentations composed of relevant
pieces of (multimedia) content.  By discarding this simplifying
assumption we need to provide a
replacement which allows the expression and resolution of these interdependencies.


Our approach thus leads to two different but related scientific challenges:

- to realise a paradigm shift where document engineering is enhanced
to the extent that it is able to steer the presentation creation
process;

- to provide a sound scientific basis to elicit knowledge from
existing communication theory.

To achieve general and sound results, we hypothesize that

1) effective presentation is only possible to the extent that one
   can formally describe and re-use the context of information delivery

2) the formalism should accommodate ambiguous, incomplete and
   contradictory viewpoints

3) explicit modelling is required to provide a sound basis,

4) progress in this area requires a tightly coupled theory-experiment cycle.


An initial requirement for communicating information is that the
relationships among the underlying domain knowledge are known.
Expressing this domain knowledge for computer manipulation is being
tackled by many other research groups and this proposal builds on
their results.  A domain model by itself is, however, insufficient.

Part of the research is to identify the other types of knowledge we need
to capture.  A system for conveying information effectively requires
knowledge about how users understand information in presentations.  Our
sub goals are thus to capture knowledge on different types of human
communication that can be applied to effective communication of
information.  Explicit models of discourse and graphic design can be
used to create better presentations in general.  These also need to be
guided by user-specific requirements and preferences.  Examples of
three types of knowledge that need to be made explicit are the
following:

Discourse - theories of communication (for example Grice's maxims)
state that information should be relevant, correct, sufficient and not
excessive.  

Graphic design - Information visualisation is a key element to convey
information. Ranging from simple charts to intricate 3-D VR
environments.  

User model - Often this is a reflection of the knowledge of
the subject-matter domain but needs to be extended to capture
preferences in terms of discourse and graphic design.


The information revolution we are currently part of will not be
complete before we have grasped the science to make full use of the
computer's potential as a comprehensive and intelligent interface to
massive amounts of information.

-----