Probabilistic Reasoning

In classical logic it is usually assumed that a proposition is either true or false, whereas in real life the truth value of a proposition may be known only with a certain probability or not at all. Since such uncertainties are often encountered in the field of expert systems, the knowledge representation and especially the problem-solving strategy must be extended accordingly. The two basic approaches are probabilistic reasoning and non-monotonic reasoning, which is discussed in the next chapter. The basis of probabilistic reasoning [Genesereth 87] is the attachment of a probability to each proposition to express the uncertainty. The uncertainties may be derived from representative statistics or estimated by experts.1