ABSTRACT
This chapter focuses on one assistance property: the ability to handle inexact knowledge and instructions. Research on this point was done at GMD within the TASSO project. The basic motivation for studying and applying nonclassical inference methods lies in the strong need for flexibility in order to handle inexact knowledge. A first concrete result toward the goals described is the Associative Memory Model of Henne, a flexible experimental system that tackles different problems at different levels. The level offers various forms of associative memory structures like object-attributevalue triples or chains of predicates. The main advantage here is that fields of the same size are easily translated into an associative model with fixed input and output vector. The study of nonmonotonic inference mechanisms is of central interest because they allow the assisting system to handle incomplete problem specifications on the basis of standard assumptions.
