Trip Report

Fraunhofer - Institut Angewandte Informationstechnik

26/02/2004 - 27/02/2004

by Stefano Bocconi

The trip was organized with two intents: see whether there could be a fruitful collaboration with the group led by Leonie Schafer and gain some narrative knowledge from people working in that field.

About my work

To show what my work was about I gave a presentation, which basically illustrate a mechanism to create biased documentaries (the subject of the ACM MM04 paper submission). One of the questions I got about the presentation was why would you do something like this, i.e. about the motivation, indicating the fact that the need for automatic generation of presentations is not self-evident. This was unexpected, because I thought that the focus of the group was in narrative and automatic narrative generation. As I saw later, that is only one part of the group activity and not everybody is working on that (actually only 1-2 persons). Another remark I got uncovers a known weak point of my approach, which is how do I know what is the impact on the user of the generated presentation, especially when dealing with media items meant to elicit emotions (we have to assume there is a user model that tells us the user's response).

Other questions they asked were:

What kind of logic do you use? (I thought First Order Logic, but at the moment I use a sort of home made logic that suffices)
Could you do it automatically based on audio transcript? (Well, only to a certain extent, but this is something interesting to investigate with a possible collaboration with Streamsage)

About their work

I got a guided tour in the Department; I saw three diffferent activities:

Augmented Reality

This is ad environment where real objects and virtual ones are mixed so that they both influence each other behavior (for ex. the virtual agent can not go through the real closed door). There are two kind of this environment, with humans and without them. An application of the first kind is the Augmented Round Table for Architecture and Urban Planning, where real designer use real objects (tracked by the systems) togheter with computer supplied visualisation. An application of the second kind is Mixed Reality Stage, where a music show is simulated with real puppets and props together with virtual ones to aid a show designer to see the end result without having to implement it.

Computer Aided Collaborative Work

This institute has developed the famous Basic Support for Collaborative Work (BSCW), a 'shared workspace' system over the Web which supports document upload, event notification, group management and is accessible with a standard Web browser. They build applications on top of that like Basic support for Collaborative Learning (BSCL) and TOWER. This last one provides a visual representation of collaborative activities of team members and their shared working context in a Theatre of Work, and it is meant to summarize and give overview of work done by distant member of a team.

Narrative

Narrative is more of an interest for Leonie (conflicting in resource with her responsability as department chief) and not much work has been done yet. She has the idea of automaticaaly generate a story using pictures, from an example of a famous german photographer published on some german newspaper (I do not remember the nemes :-( ).

The collaboration

We decided to write together a paper about collaborative language learning for children: children from different nationalities would create stories with pictures and annotate them in their own language. Other children would see the stories (possibly generated by the system rearranging the original stories) and learn the language associating the annotations (language dependent) with the pictures (language independent).
The idea is nice but there are a lot of hard-to-solve issues, so this paper will be a first small step. My contribution should be on the output side (SMIL).