1. Enhancement of the theoretical basis
   Despite its long tradition, work on foundational issues in many-valued logics still spawns new and thrilling problems and results. The computational properties of these logics are still not completely understood. In particular the investigations into normal forms pose a remarkable challenge with respect of the required mathematical tools. Also the algebraic and relational representation of many valued logics needs further clarification. But probably the most inter esting issue is the exact relationship both in theory and in implementations between fuzzy logic, using infinitely many truth values, and its finitely valued approximation. Progress in this topic may also have a substantial impact on applications.
2. Automated deduction: theory and tools
   Previous research already showed that many valued reasoning may in principle be mecha nized. Recently, inference systems geared towards machine implementation emerged and have already been realized in experimental automated reasoning programs. To name just two of the more recent approaches we mention labelled deductive systems and inference methods based on linear and integer programming.
3. Modelling of and reasoning on incomplete and uncertain knowledge
   Many-valued calculi are often encountered, for instance, in possibility theory and possibilistic logic. They provide a natural framework for manipulating orderings between preferred inter pretations, or more or less entrenched propositions in nonmonotonic reasoning. Still another example is given by conditional objects which are symbolic counterparts of conditional prob abilities and are 3-valued logical entities. Besides, truth-functional fuzzy logic has proved to be useful for modelling interpolative reasoning, as used in fuzzy control applications for in stance.
4. Applications
   The main thrust in applying many valued logics lies in fuzzy logic applications for intelligent and adaptive control systems and has already been incorporated into some innovative indus trial products. The current situation is characterized by:
• developing an application culture using existing tools and methodologies,
• establishing tool kits as a core technology platform for industry,
• exploring further improvements and extensions of fuzzy logic technologies, e.g. in combi nation with neural computing to mention just one example.

Connections between modal and many valued logics also show fruitful results in modelling and reasoning about intelligent multi-agent systems.

A very interesting approach, finally, is the application of three valued logic to adaptive error- correcting communication, where the third truth value, besides true and false, designates the possibility of a transmission error.

A great number of the expected participants of this seminar are members of the COST action # 15 on ,Computer Applications of Many-Valued Logics". The Dagstuhl seminar will constitute the annual COST workshop conference and will offer the opportunity to present the work done within this action to a worldwide audience.