Trip Report on visit to GMD-IPSI, Darmstadt Date: Mon, 30 May 1994 20:07:00 +0200 By: Lynda Hardman Report on trip to GMD-IPSI, April 25th - May 17th 1994 ERCIM Human Mobility Programme Introduction I chose to visit GMD-IPSI because I have come across their work in the literature and have been following other, more related, work via the GMD technical reports. My proposed PhD work lies extremely close to work going on in the PAVE group there. I had two goals for my visit: to take a month out of the normal office "drudgery" to concentrate on my PhD work, and to get to know the different groups at IPSI and see where connections could be made between them and CWI. While I was there I gave two talks on my work (Amsterdam Hypermedia Model and Structured Multimedia Authoring) and got some lively feedback. I think it would be very useful to go back around September/October time (for a week or two) to see how the people I talked to have progressed with their work, and to discuss where my PhD work is. This report is fairly long, but structured, so feel free to skip bits you aren't interested in: Overview of GMD-IPSI, along with short descriptions of some of the groups. More detailed description of PaVE (my host group). Very short note on my PhD work. Notes on possible future contacts. ----- GMD-IPSI GMD as a whole has 8 sites in Germany, with 1200 employees. 4 sites are in Bonn, 2 in Berlin and 2 in Darmstadt. IPSI is one of the 2 in Darmstadt and has 90 employees. IPSI is the Integrated Publishing Systems Institute. Professor Dr Erich Neuhold is the director of GMD-IPSI. His interests are in databases. His deputy is Dr Dr Norbert Streitz, whom I know better because of our common hypertext interests, although his lie very much in the CSCW direction. IPSI has 4 research groups, plus an applied group. The 4 research groups are (and I quote): Datenbank-/ Informationssysteme (ooDBMS, Integration, Multimedia) Mensch-Computer-Interaktion (Information Retrieval, Visualisierung) Verarbeitung natuerlicher Sprache (Textanalyse und Textgenerierung) Kooperative Hypermediasysteme (Hypermedia-Publikationen, CSCW) The applied group is Experimentierumgebungen fuer MM-Produktie (MultiMedia-Forum) IPSI is about half the size of CWI, with more people working in less spread-out fields. I visited the PaVE group who are part of Hypermedia publications. I also got to know people working in CSCW - the WIBAS group - quite well. The Multimedia Forum are putting together information about the work going on here and making it available via WWW. The URL is: http://www.darmstadt.gmd.de/gmdda.ip.html ----- Matthias Hemmje's group During my visit there was an informal multimedia workshop with some people from IPSI and Simon Gibbs from Geneva. Simon gave a short talk, and from what I could gather, he starts off with extremely similar research goals to those of the Multimedia group at CWI (e.g. can map to different piece of hardware/software, synchronize different sources pf media and virtual machine mapped to different architectures) and comes up with entirely different solutions. He has a more object/method type approach where different bits can be slotted together with each other. Simon is familiar with Dick Bulterman's work. Perhaps more interesting is the fact that Simon will be moving to GMD, Bonn, at the end of the summer. He'll be setting up a multimedia lab (of some sorts) there. Apparently there already is a 140 MBper sec network connecting up different places but it costs 60 DM per minute to use (within Germany), so not many people are using it right now. Bonn should be one of the ATM sites. Two of the talks were "The virtual studio" and "DIVIDED: The DIstributed VIDeo EDiting system" (I have copies of the slides for the latter) both of which fall under "The Integrated Video Processing Environment". I get the impression that the guys here are at the stage with video that we were about 2 years ago. (Although they are now starting with more standard technology.) Virtual Reality - This was also the group Monica Bordegoni (ex-CWI ERCIM fellow) worked in during her stay here. She put together a set of gestures for their data glove, but they don't use it any more because they can't get sufficient accuracy out of it to make it useful for pointing in their applications. I did see a demo of a database query system with a display of documents you could choose from and the results of your choice were displayed using 3D graphics. ---- VODAK Thomas Rakow is the group leader working on multimedia databases and information retrieval. There were a number of interesting reports (which I passed on to Carel van den Berg in AA). There is also a shortpiece in ERCIM News No. 17, April 94, pp 26- 27. ----- ----- PaVE PaVE (Publication and Visualisation Environment) is the group I was visiting. It is headed by Christoph Huser. Thomas Kamps was my host. Thomas is working on automatic layout specifically for querying an art encyclopedia. He works closely with Klaus Reichenberger, a graphics designer. There are others working on building up the information underlying the encyclopedia. They are converting typesetter marked-up information to SGML (and have tools to help the author as much as possible). There are 7 in the group (including Christoph). Projects are DoA (Dictionary of Art, in full swing), IEN (Individualised Electronic Newspaper, ended), DPA (a project with Reuters). The members of the group are working on different parts of the different projects. Expertise is split into smalltalk programming, graphic design, information retrieval, indexing document structures, and automatic layout generation. There are two guys mainly concerned with the Smalltalk environment (Lothar Rostek and Dietrich Fischer). Thomas is designing and implementing auto graphic layouts for the Dictionary of Art, Klaus does user interface design. Wiebe Moehr is into document structures. Anja Haake has just finished her PhD thesis on versioning in hypertext. The group programs everything in Smalltalk, using a home-grown environment SFK - Smalltalk Framework Kit. (They used to use HyperNews, but gave up on that a year or so ago.) My impression is that they like Smalltalk for the same sorts of reasons that we like Python. They also have the advantage that Smalltalk runs on multiple platforms and is becoming more and more popular in the commercial world. (BMW have just bought 2000 licenses and HP are going to distribute Smalltalk on their machines.) SFK is available to any (academic) who wants it. I had one long chat with Christoph Heuser, the group leader. (Unfortunately he left for a long holiday, so I only get a chance to talk to him between the going-away panics in the first week.) He has done a lot of the work on automatic layout in their previous project IEN (Individualised Electronic Newspaper) and gave me an overview of how far they got and what all the pitfalls are. His and my vision on what online information is/should be are very similar. They have various good contacts with PIRA/Roberto Minio (they are a British, originally paper-based, printing house looking at the new technologies - I've been impressed by Roberto Minio in the past ) and MacMillan - publishers. They are also starting a project with Reuters (but there will be a lot of heavy duty database work which they are hoping will be done by a software company). Their research is very much text-based, and although they too want distributable documents over a number of platforms, they aren't focussing on the multimedia issues. Also their approach is very much "knowledge-based". They are working up from the domain semantics and would really like to generate the text rather than use pre-created texts. In the meantime they are using a highly structured database of SGML documents with a domain index. This makes our research interests very close by but not competing. ----- WIBAS group, Dolphin project Norbert Streitz's main interests are in the CSCW work, where they are looking specifically at using hypertext for supporting cooperative work. Joerg Haake is the group head. Part of the work of the WIBAS group is the Dolphin project. This is an environment for cooperative work - in this case a meeting room. There is a large (computer) screen you can write on with a pen and then you sit in front of your own screen with the same communal image (of the big screen) and have also your own private workspace. They are at the stage of building the tools and want to try out some real sessions with 6 people. To be honest I can't quite see the added value in this yet. They aren't going for remote meetings - and even if they were, there is no audio, let alone video, support. It's sort of nice to have a computer screen to write on instead of a whiteboard, but I don't see the added value of having your own workstation (in the meeting environment). Anyway, they might be a fun group to try out some remote UniSim Deluxe stuff with. (They are not one of the first sites to get ATM connections, which is a shame.) It would also be interesting to have Gloria Mark, a psychologist who has just joined them, come up and visit us some time. Gloria is doing some testing with the liveboard system (basically a computer screen the size of a blackboard) they have set up here. The idea is that you can have a meeting where the participants can explore their ideas and create a hypertext structure to illustrate their ideas, and maybe even the argument structure. It was interesting trying out the system. I got to use the liveboard in a first private session. It's a bit like being handicapped in the use of your computer - although you are using a pen, rather than a mouse, the writing is worse than with chalk on a blackboard, and you have to press hard or the system thinks you've lifted the pen off. To increase "usability" you can also use a keyboard - and then you type looking at the small keys in front of you and you have no idea what is appearing on the board. The gesture interface was interesting however. As you scribble, the interface is checking whether you have made any known gestures. In the different modes of operation only a few gestures are distinguished. Unfortunately sometimes two of the gestures are fairly close (in syntax) and the system gets confused about which gesture you meant. The actual experiment should've taken two hours. Instead we spent the whole day trying to use the software, giving up because it was too buggy, taking breaks and trying later. I have now learned that you should **not** give guinea-pig users software that you haven't tried out for a long time. Even just one fall into the debugger makes the user wonder what the ***** they are doing there and what they did wrong, and really disturbs the task in hand..... ------ ------ My PhD work I spent a number of sessions talking to Thomas and Klaus about their automatic layout work. Thomas is a mathematician trying to find the right graph analysis procedures for finding regularities for his layouts. Klaus is a graphics designer trying to express what it is he does in terms that can be automated. I took lots of notes on my discussions, and even wrote them up hastily and had Thomas and Klaus comment on them, so at least I have some record of what they told me. The problem they want to solve (graphically submitting and editing queries) is subtley different from mine (layout of multimedia objects), but there are mutual problems that need to be solved, so it was extremely useful talking to them. ----- Possibilities for future contact There's a good chance that Anja Haake (PaVE) and Joerg Haake (WIBAS) will come and visit us in Amsterdam sometime in the summer. We should invite Christoph Hueser to give a CWI seminar. He could give an interesting overview of their approach(es) and some examples of what they have done. It would be interesting to have Gloria Mark visit and discuss the user aspects of the multimedia meeting work. Another visitor from Canada, Gene Golovchinsky, has just done his masters and is starting a PhD in something which sounds pretty much the same as the line I'm on. He's also visiting IPSI to talk to Thomas and Klaus about layout, and then think about auto multimedia layout. His plan is to take back the Smalltalk programming environment - SFK. His supervisor, Mark Chignell, has some contacts with IBM and they are looking into multimedia editors, so Gene is also interested in CMIFed. I would still like to speak to people: PavE: Christoph Heuser, Joerg Geissler, Anja Haake; Multimedia Forum. -----