<XML><RECORDS><RECORD><REFERENCE_TYPE>10</REFERENCE_TYPE><REFNUM>5966</REFNUM><AUTHORS><AUTHOR>Johnson,C.W.</AUTHOR></AUTHORS><YEAR>2001</YEAR><TITLE>Models and the use of Counter-Factual Reasoning in AccidentInvestigations</TITLE><PLACE_PUBLISHED>DCS Tech Report</PLACE_PUBLISHED><PUBLISHER>Dept of Computing Science, University of Glasgow</PUBLISHER><LABEL>Johnson:2001:5966</LABEL><ABSTRACT> The increasing complexity of many accidents and incidents poses considerable challenges to the investigators who must distinguish causal factors from a mass of other contextual details and contributory factors. The analysis of failures involving interactive systems, typically, relies upon multi-disciplinary teams who may also form very different opinions about the causes of an incident or accident. In consequence, it can be difficult to ensure that any resulting analysis is consistent with the available evidence and that it does not overlook other important information about the course of a failure. This paper argues that mathematically based techniques can be use to address these problems. In particular, we examine both deterministic and stochastic models of causation. The lack of ?real world? experience in the application of these innovative techniques places considerable constraints on the conclusions that can be drawn from this preliminary work. It is, however, possible to identify opportunities for the pragmatic application of these theoretical techniques to address the growing problems of accident and incident investigation for complex, interactive systems. </ABSTRACT></RECORD></RECORDS></XML>