EVE: a framework for event detection

In this paper, we introduce EVE, a generic framework for event detection where events can also include outliers, model changes and drifts. Various methods for event detection have been proposed for different types of events. However, many of them make the same or very similar prior assumptions but use different notations and formalizations. EVE provides a general framework for event detection, which allows existing algorithms to be represented using a common basis. The framework includes generic types of time slots, streaming progresses, and measures of similarity between those slots. We demonstrate how existing algorithms fit nicely into this framework by instantiating appropriate window combinations, progress mechanisms, and similarity functions.