Abstract
The main aim of this paper is to present a way to improve the efficiency of an intelligent alarm correlation module by means of an abductive reasoner. The alarm correlation module is integrated in a more complex system (SAMOT) that performs monitoring and control tasks over a tract of a highway. On the basis of a specific theory on the domain, explanations of anomalous traffic patterns can be provided taking into account those situations not directly detected by data acquisition technology. The integration of the comprehensive view obtained by means of a correlation phase with additional domain knowledge allows to abduce from the observed anomalous traffic patterns other significant situations. The experience of traffic operators allows to build a theory of abducible explanations that exploits the knowledge of the use context of the system. The correlation module of SAMOT has been modelled within a propositional modal language, named ST-Logic, which will be adopted for the abductive reasoner as well. Finally, some remarks about stratified interpretative cycles made possible within such a formal framework will be addressed.