Using Context Information for Efficient Situation Detection

In contrast to commercial trigger databases, active databases support complex situation detection. In some active database applications, those situations are not restricted to transaction units, but may span the whole lifetime of the database application. As a consequence, the amount of process information cumulates within the system and may reach problematic dimensions. This paper introduces a method to restrict the amount of events and potential complex situations by deriving more and more precise constraints from significant context information during the collection of indicators for a complex situation.This zoom effect is a natural way to model the continuous narrowing of relevance, as known in cognitive science. Our approach extends the parameterization of events in related research. A combination with established restriction techniques significantly restricts both relevant basic events and complex situations. This leads to an efficient situation detection even for long running activities.