Learning is obviously an important phenomenon of natural intelligence. Therefore, despite diverging specifications of the area of artificial intelligence, learning is a central issue of artificial intelligence research. There is abundant evidence of the human ability to learn from possibly incomplete information. In human communication, one usually provides only incomplete information with respect to some target phenomenon to be described or specified, or even to be learned. Algorithmic or computational learning theory is the theoretical division of machine learning research dealt with those problems. Analogical reasoning and inductive inference are typical research areas faced to the central problem of processing possibly incomplete information.

Nowadays, there are three international conference series in the area of algorithmic or computational learning theory. The youngest one is ALT (Algorithmic Learning Theory) established in Japan in 1990 and held annually since then. Two years earlier, there has been started the annual workshop series COLT (Computational Learning Theory) in the USA. Already in 1986, a workshop series called AII (Analogical and Inductive Inference) has been established in Germany. Compared to ALT and COLT, the AII events are distinguished by a considerably smaller number of participants as well as by putting much more emphasis on longer talks and room for discussions. So, the International Research and Conference Center at Dagstuhl Castle seemed particularly tailored to host AII'92. The organizers gratefully acknowledge the excellent working conditions provided for this third event in the All series.

The conference programm contains presentations of different type. First, there are invited talks by distinguished scientists reporting on essential contributions to the area. Second, there are submitted papers which have been processed in a reviewing and selection process as usual. Third, there have been invited two talks of a very special type to be presented and discussed at evening sessions. Fourth, according to the policy of Dagstuhl Seminars to provide every participants the opportunity to present her/ his recent work, there are so-called supplementary talks. The programm attached at the end of the Dagstuhl-Seminar-Report allows to get an overview of these categories of contributions. The invited talks and the submitted papers which passed the reviewing process successfully are published in the proceedings volume Analogical and Inductive Inference, Klaus P. Jantke (ed.), Lecture Notes in Artificial Intelligence, Springer-Verlag, 1992. Readers interested in the proceedings of the former AII workshops may consult Springer-Verlag Lecture Notes in Computer Science 265 and Lecture Notes in Artificial Intelligence 397. Selected contributions will appear in a revised form in a special issue of the Journal of Experimental and Theoretical Artificial Intelligence (JETAI). This will be volume 5 (1993), issue 2.

As usually behind the scene, the work of a number of colleagues contributed essentially to the success of the workshop. I want to express my sincere gratitude to all of them. The Algorithmic Learning Group of Leipzig University of Technology, in particular Andreas Albrecht, Ulf Goldammer, Steffen Lange, and Eberhard Pippig, provided continous assistance. Steffen Lange did a particularly important Work as organizing secretary of AII'92. Last but not least, Thomas Zeugmann (Darmstadt) has intensively pushed forward the assembly of the present publication.

Creative Commons BY 3.0 DE