The KI-2003 Proceedings (Springer LNCS 2821) are available online.
You can find it here.
The KI-2004 will be held in Ulm on September 20-24 next year..
Invitation to KI-2003 in Hamburg Bernd Neumann, Conference Chair
The programme of the 26th German Conference on Artificial Intelligence has been established and we look forward to an exciting conference with invited talks, refereed papers, posters, workshops, an exhibition and, of course, attractive social activities in the beautiful town of Hamburg. We invite everybody to come to Hamburg and enjoy this conference!
The German Conference on Artificial Intelligence is a traditional and unique yearly event which brings together the German AI community and an increasing number of international guests. While not as old as IJCAI (which first took place in 1969), KI-2003 marks a tradition which officially began 1975 with a workshop of the Working Group "Künstliche Intelligenz" of GI. Actually, there was one important AI conference in Germany before this, the "Fachtagung Cognitive Verfahren und Systeme" (Cognitive Methods and Systems) held in Hamburg in April 1973. It is a special pleasure that one of the organizers of the 1973 event, Hans-Hellmut Nagel, will also participate in KI-2003.
Invited Talks
KI-2003 features five invited talks by outstanding leaders of the field, addressing advanced work on important AI topics. The speakers are Nick Jennings (University of Southampton), Daniel Keim (University of Konstanz), Erik Sandewall (University of Linköping), Rudi Studer (University of Karlsruhe) and Wolfgang Wahlster (DFKI Saarbrücken). We believe that their presentations will provide insights valuable for any AI researcher attending the conference.
Workshops
Looking at the programme of KI-2003 we see 11 workshops on the first two days of the conference. The workshop themes cover diverse active research topics and exemplify important developments of the field. Some of the workshops have there own tradition as a workshop series and are co-located with the main conference. It has been a deliberate decision of the organizers of KI-2003 to encourage workshop activities within the overall conference programme. One reason is to offer platforms for the presentation of early results and for specialized discussions. As AI has split into more and more diverse subfields, there is an increased need for such specialized subevents. Another reason is to draw more people to the conference than could be expected from the main conference program alone. It is the fate of many national AI conferences that papers are preferably submitted to international AI conferences, hence participation is an issue.
Papers and Posters
There have been 90 submissions from 22 countries. 18 papers have been accepted for oral presentation and 24 papers for poster presentation. The acceptance criteria have been set to meet high international standards, so the resulting programme can be expected to offer first-class research results. Papers will have 30 min for oral presentation. Posters will be introduced by short spotlights (a first at KI conferences) and presented in a special poster session at prime time. Papers as well as posters are allotted 15 pages each in the proceedings.
The topics of the sessions are representative of several important developments of the field. As a first highlight, I want to mention work on multimodal information processing. Multimodal aspects are addressed in several contributions, for example, on information fusion, vision-language integration, dialogue control with integrated gesture and facial expression analysis. Furthermore, the opening plenary talk by Wolfgang Wahlster will address this topic head-on. The interest in multimodal information processing is a positive indicator of integration efforts across subfield boundaries.
Another interesting development is the integration of cognitive modelling with AI engineering tasks. Advanced user interfaces provide an important motivation, as human cognitive mechanisms and constraints have to be considered when shaping human-computer interactions. The interest in cognitive modelling may also reflect a certain amount of frustration about more formal approaches to human-type reasoning. More and more advanced applications - for example in robotics, decision making, high-level vision, diagnosis, planning - ask for some sort of common-sense integration which is still difficult to provide in a strictly formal framework. Hence high-level cognitive empiricism enjoys a revival.
Work on integrating perception and symbolic reasoning is yet another area of interest. While internationally the respective research comunities are still largely disjoint, there exists a long history in Germany of research combining the two fields. Also, several EU-funded projects devoted to "Cognitive Vision" have been initialized in the last few years, hence the development can be expected to intensify. The main integration challenge for vision and perception research seems to be the incorporation of logic-based and qualitative reasoning mechanisms as a top-down support for signal interpretation. On the other hand, AI researchers have to adapt reasoning mechanisms for hypothesis generation and uncertainty management as required for vision and perception. It is worth noting that AI and Computer Vision have not beern separated initially. The interested reader is encouraged to (re)read Raj Reddy´s paper "Eyes and Ears for Computers - A Review" in the proceedings of the above mentioned 1973 conference in Hamburg. At that time, a coherent, unified view was still possible.
Message of a Senior AI Scientist
Being the Conference Chair of KI-2003, the past Programme Chair of GWAI-83 20 years ago and a participant of the 1973 conference in Hamburg, I have the privilege of looking back at AI conferences and AI research of four decades. From the many experiences which unavoidably remain, I want to shortly address the experiences connected with the desintegration of the field into subfields and separate scientific communities, as already commented on earlier. Clearly, as a field grows, diversification and specialization is necessary as a sort of divide-and-conquer strategy. This is no problem as long as the overall context in terms of long-term research goals and visionary applications remains visible and outspoken. Unfortunately, as specialized research communities evolve, there is often a tendency to establish a self-contained research culture with little or no attention to goals beyond one´s own field. Blame has to be put primarily to the award system for scientific research which favours intracultural work and is of disadvantage for intercultural work.
As responsible scientists, we should counteract these tendencies. First, we should keep asking for the long-term goals of our research. One useful perspective is the vision of a future application scenario. If I cannot think of a long-term application of my work, I would mistrust its usefulness. Another recommendation is to deliberately cross interdisciplinary boarders. This is not only necessary to develop an integrated view, but may also shed new light onto one´s own discipline. On an occasion like this conference, we may also have the chance to influence the award system. It has been my goal as a conference chair to not only judge papers for their rank in a scientific competition with rules established by the experts of each respective subfield, but also for their merit as a contribution in a larger context. I trust that this view point is to the better of our AI conferences.
