A framework for event pattern recognition

Abstract

Complex Event Processing is the ability to interpret and handle the complex characteristics defined in events, occurring in the world of business activities. This emerging technology is a breakthrough in the world of event processing. It enables IT professionals to understand the full environment of a system. This brings to light possibilities where the performance of a system can be improved through the manipulation of related events which form patterns. This dissertation discusses how the complexities in systems can be solved through the organisation of the events taking place. In this way, we present a framework which aims at achieving a flexible and better performance for event pattern recognition. We introduce a method to provide an efficient and scalable way of interpreting such sought after activities in the environment. Furthermore, we introduce an optimised way of filtering the ever complex world of events occurring in the environment. This helps in increasing the performance rate of the system, by retrieving the most relevant of events to be considered for event pattern recognition. We implement an Event Pattern Language, with the purpose of enabling the user to define these sets of desired patterns. The pattern analysis process is handled transparently by the framework. In the background, state machines track the currently matched events of a particular pattern introducing incremental gain on performance rates. State machines also perform on-the-fly pattern analysis of the dynamic properties of a pattern which in tum require runtime (real-time) evaluation.