ABSTRACT
This chapter provides a framework in which the operative conditions of existing buildings can be assessed on a probabilistic basis. The decision maker system is able to learn from the expertise it collects during its service. The supporting mathematical tool makes use of a causal probabilistic network to represent the knowledge domain. One can try to use it in a dynamic process by which the status of uncertain knowledge is updated on the basis of the collection of new data. The probabilities associated with variables for which a direct observation is still lacking should be modified as new pieces of evidence arrive. BAIES is an experimental computer code for the analysis of Bayesian networks of discrete variables. It is a probabilistic expert system for computing the marginal probabilities of oriented graphs whose nodes represent discrete random variables: the properties of conditional independence are captured in the topological structure of the acyclic oriented graphic.
