Problems in decision making and in other areas such as pattern recogni-
tion, control, structural engineering etc. involve numerous aspects of
uncertainty. Additional vagueness is introduced as models become more
complex but not necessarily more meaningful by the added details. During
the last two decades one has become more and more aware of the fact that
not all this uncertainty is of stochastic (random) cha- racter and that,
therefore, it can not be modelled appropriately by probability theory.
This becomes the more obvious the more we want to represent formally
human knowledge. As far as uncertain data are concerned, we have neither
instru- ments nor reasoning at our disposal as well defined and
unquestionable as those used in the probability theory. This almost
infallible do- main is the result of a tremendous work by the whole
scientific world. But when measures are dubious, bad or no longer
possible and when we really have to make use of the richness of human
reasoning in its variety, then the theories dealing with the treatment
of uncertainty, some quite new and other ones older, provide the
required complement, and fill in the gap left in the field of knowledge
representation. Nowadays, various theories are widely used: fuzzy sets,
belief function, the convenient associations between probability and
fuzzines etc --- We are more and more in need of a wide range of
instruments and theories to build models that are more and more adapted
to the most complex systems.