The relationship between belief networks and relational databases is examined. Based on this analysis, a method to construct belief networks automatically from statistical relational data is proposed. A comparison between our method and other methods shows that our method has several advantages when generalization or prediction is deeded.

In mechanical design, there is often unavoidable uncertainty in estimates of design performance. Evaluation of design alternatives requires consideration of the impact of this uncertainty. Expert heuristics embody assumptions regarding the designer's attitude towards risk and uncertainty that might be reasonable in most cases but inaccurate in others. We present a technique to allow designers to incorporate their own unique attitude towards uncertainty as opposed to those assumed by the domain expert's rules. The general approach is to eliminate aspects of heuristic rules which directly or indirectly include assumptions regarding the user's attitude towards risk, and replace them with explicit, user-specified probabilistic multi attribute utility and probability distribution functions. We illustrate the method in a system for material selection for automobile bumpers.

The presence of latent variables can greatly complicate inferences about causal relations between measured variables from statistical data. In many cases, the presence of latent variables makes it impossible to determine for two measured variables A and B, whether A causes B, B causes A, or there is some common cause. In this paper I present several theorems that state conditions under which it is possible to reliably infer the causal relation between two measured variables, regardless of whether latent variables are acting or not.

In probabilistic logic entailments, even moderate size problems can yield linear constraint systems with so many variables that exact methods are impractical. This difficulty can be remedied in many cases of interest by introducing a threevalued logic (true, false, and "don't care"). The three-valued approach allows the construction of "compressed" constraint systems which have the same solution sets as their two-valued counterparts, but which may involve dramatically fewer variables. PROLIFERATION OF WORLDS An entailment problem in Nilsson's (1986) probabilistic logic derives an estimate for the prior probability of one sentence (hereafter, the "target") from the priors for a set of other ("source") sentences. V is a matrix derived from an inventory of all consistent patterns of truth assignments (1 true, 0 false) for the source and target sentences.

Survey of several forms of updating, with a practical illustrative example. We study several updating (conditioning) schemes that emerge naturally from a common scenarion to provide some insights into their meaning. Updating is a subtle operation and there is no single method, no single 'good' rule. The choice of the appropriate rule must always be given due consideration. Planchet (1989) presents a mathematical survey of many rules. We focus on the practical meaning of these rules. After summarizing the several rules for conditioning, we present an illustrative example in which the various forms of conditioning can be explained.

Irrelevance-based partial MAPs are useful constructs for domain-independent explanation using belief networks. We look at two definitions for such partial MAPs, and prove important properties that are useful in designing algorithms for computing them effectively. We make use of these properties in modifying our standard MAP best-first algorithm, so as to handle irrelevance-based partial MAPs.

This paper proposes a new method for solving Bayesian decision problems. The method consists of representing a Bayesian decision problem as a valuation-based system and applying a fusion algorithm for solving it. The fusion algorithm is a hybrid of local computational methods for computation of marginals of joint probability distributions and the local computational methods for discrete optimization problems.

The graphoid axioms for conditional independence, originally described by Dawid [1979], are fundamental to probabilistic reasoning [Pearl, 19881. Such axioms provide a mechanism for manipulating conditional independence assertions without resorting to their numerical definition. This paper explores a representation for independence statements using multiple undirected graphs and some simple graphical transformations. The independence statements derivable in this system are equivalent to those obtainable by the graphoid axioms. Therefore, this is a purely graphical proof technique for conditional independence.

A control strategy for expert systems is presented which is based on Shafer's Belief theory and the combination rule of Dempster. In contrast to well known strategies it is not sequentially and hypotheses-driven, but parallel and self organizing, determined by the concept of information gain. The information gain, calculated as the maximal difference between the actual evidence distribution in the knowledge base and the potential evidence determines each consultation step. Hierarchically structured knowledge is an important representation form and experts even use several hierarchies in parallel for constituting their knowledge. Hence the control strategy is applied to a layered set of distinct hierarchies. Depending on the actual data one of these hierarchies is chosen by the control strategy for the next step in the reasoning process. Provided the actual data are well matched to the structure of one hierarchy, this hierarchy remains selected for a longer consultation time. If no good match can be achieved, a switch from the actual hierarchy to a competing one will result, very similar to the phenomenon of restructuring in problem solving tasks. Up to now the control strategy is restricted to multi hierarchical knowledge bases with disjunct hierarchies. It is implemented in the expert system IBIG (inference by information gain), being presently applied to acquired speech disorders (aphasia).

In this article we present two ways of structuring bodies of evidence, which allow us to reduce the complexity of the operations usually performed in the framework of evidence theory. The first structure just partitions the focal elements in a body of evidence by their cardinality. With this structure we are able to reduce the complexity on the calculation of the belief functions Bel, Pl, and Q. The other structure proposed here, the Hierarchical Trees, permits us to reduce the complexity of the calculation of Bel, Pl, and Q, as well as of the Dempster's rule of combination in relation to the brute-force algorithm. Both these structures do not require the generation of all the subsets of the reference domain.