The number of publications on the management of uncertainty in expert systems has grown considerably
over the last few years. Yet the discussion is far from drawing to a close. Again and again
new suggestions have been made for the characterization and combination of uncertain information
in expert systems. None of these proposals has been adopted generally.
Most of the methods recommended introduce new concepts which are not founded on classical
probability theory. This book, however, written by statisticians, investigates the possibility of
giving a systematic treatment using the classical theory. It also takes into account that in many
expert systems the available information is too weak to produce reliable point estimates for probability
values. Therefore the handling of interval-valued probabilities is one of the main goals of
this book.