Causal Structures: From Logical Models to Diagnostic Applications

The phenomenon of causality is omnipresent in language, philosophy and science and it seems to be a core, fundamental idea in common-sense thinking. In this paper the nature of causality is investigated and some characteristics of this concept are presented. Particular attention is also paid to applications in diagnostic reasoning. Causality provides bases for diagnostic inference and explanation of failures through search in the direction against causal influences. From logical point of view, this can be considered as abductive inference. A core, generic model of the search space is built as an AND/OR/NOT causal graph speciffing the structure that is used to establish diaposes. Such a graph constitutes a schematic representation of causal influences among symptoms and incorporates logic-like functions for combining influences of multiple symptoms. An illustrative example is presented to clarify most of the presented ideas.