Where are we now?

Lynda Hardman, April 2005

We have been working on the generation of hypermedia presentations for a number of years now. First evidence of the work can be found as Semi-Automatic Generation of Hypermedia Presentations: a starting point. So, what do we have, where are the gaps, what should we do in the short term and where should we aim in the long term?

What do we have?

• We can generate presentations that make trade-offs in space and time Joost's MMM01 paper.
• We are beginning to understand what we need to express in discourse and how to express it DISC and Vox Populi at MM04 SRCM workshop and ICME), plus the work done in a number of leesklubs. Personally I think this is worth some polishing and then publishing.
• Joost's work is getting us closer to understanding what needs to be included in an architecture for dealing with all the different required types of input.
• Frank and Jacco's IEEE-MM "obscure object of desire" papers and Joost's work recently on media annotation schemas and on Media Streams is shedding light on what exactly it is that makes media semantics different from text semantics.
• Lloyd has been working on the domain-independent side of things.
• Michiel Kauw-A-Tjoe did some nice work on capturing 2D abstract graphics design rules. A particular instance of design semantics.

What don't we have?

• Rob started work on trade-offs between space/time and links, but somehow never managed to solve the problem.
• Amit did some excellent preliminary work on representing colour and including it in the Cuypers engine, but, again, this didn't get taken to fruition. We still need a colour ontology and colour rules.
• I would still love to do more serious work on capturing more design semantics for inclusion in the generation process.
• We know that modalities are a large problem, but we still need to do some serious work on them. I keep thinking that capturing the semantics implicit in (a selected subset of) WCAG 2.0 and EARL (activity about to be revitalised) could be a useful starting point.
• We know the user model has to be taken into account, but have no real demonstrations yet of how. Suzanne Little has done preliminary work; Susanne Loeber has a really rich model of user characteristics, but we can't quite get at it (yet); Katya has a user model somewhere in SampLe, but it is not (yet) a realistic model of a real user.

What should we do next?

I would appreciate discussion on this.

$Id: whatNext-apr2005.html,v 1.3 2005/05/13 06:17:11 lynda Exp $