Qualitative Propagation and Scenario-based Explanation of Probabilistic Reasoning
This work addresses the need for explainable Bayesian methods in expert and decision support systems, offering incremental improvements over existing qualitative approaches.
The paper tackles the problem of making probabilistic reasoning more comprehensible by proposing two explanation strategies: qualitative belief propagation, which traces evidence effects through belief networks, and scenario-based reasoning, which generates causal stories to approximate results. The result is a framework that allows users to control explanation style and abstraction, though no concrete performance numbers are provided.
Comprehensible explanations of probabilistic reasoning are a prerequisite for wider acceptance of Bayesian methods in expert systems and decision support systems. A study of human reasoning under uncertainty suggests two different strategies for explaining probabilistic reasoning: The first, qualitative belief propagation, traces the qualitative effect of evidence through a belief network from one variable to the next. This propagation algorithm is an alternative to the graph reduction algorithms of Wellman (1988) for inference in qualitative probabilistic networks. It is based on a qualitative analysis of intercausal reasoning, which is a generalization of Pearl's "explaining away", and an alternative to Wellman's definition of qualitative synergy. The other, Scenario-based reasoning, involves the generation of alternative causal "stories" accounting for the evidence. Comparing a few of the most probable scenarios provides an approximate way to explain the results of probabilistic reasoning. Both schemes employ causal as well as probabilistic knowledge. Probabilities may be presented as phrases and/or numbers. Users can control the style, abstraction and completeness of explanations.