In this paper we propose an approach to knowledge intensive CBR, where explanations are generated from a domain model consisting partly of a semantic network and partly of a Bayesian network (BN). The BN enables learning within this domain model based on the observed data. The domain model is used to focus the retrieval and reuse of past cases, as well as the indexing when learning a new case.

Urban threat environments, including urban warfare and disaster scenarios, can be characterized as complex systems. Decision making in urban threat environments may be difficult because the underlying system exhibits non-linear and path dependent behavior that humans, unassisted by computers, have trouble understanding and reasoning about. Thus, it may be difficult to predict how the consequences of an action may unfold. VISTA, a computational model is presented as a decision-support tool to be able to help the analyst understand the complexity of an urban threat environment, thereby enabling forecasting of conditions and the exploration of potential consequences of potential actions. A demonstrative use-case is presented along with hypothetical results to illustrate the tool’s potential usefulness. Attention is also paid to the usefulness and challenges of providing an “explanation ” of the tool’s results. Initial approaches to the problem are presented. Contact:

This paper explores the role of causal explanations in evaluating counterfactual conditionals. In reasoning about what would have been the case if A had been true, the localist injunction to hold constant all the variables that causally influence whether A is true or not, is sometimes unreasonably constraining. We hypothesize that speakers may resolve this tension by including in their deliberations the question of what would explain the hypothesized truth of A. To account for our recent psychological findings about counterfactuals, an alternative approach based on Causal Bayesian networks is proposed in which the intervention operator utilizes the agent’s beliefs about the explanatory power of the antecedent of the counterfactual. The results of three psychological experiments are reported in which the new method succeeds in predicting subjects ’ responses while the traditional method for evaluating counterfactuals in Bayesian networks fails.

comparison of approaches

Abstract not found