An adaptive event stream processing environment

Abstract

With the increasing application of Event Stream Processing (ESP) for event pattern detection, it has become important to enhance the extant ESP capabilities to deal with applications having dynamic behavior. This dissertation research explores the limitations of current ESP systems due to fixed pattern detection mechanism and discusses the motivational ideas that demand enhancements in ESP. We propose a solution called adaptive ESP that explores, learns, and updates evolving patterns in dynamic applications. Development of adaptive ESP requires several research issues to be addressed: such as handling input data streams, enhancing event languages with probabilistic information, using machine learning algorithms, and processing feedback from experts. We discuss these issues with the proposed architecture for the system and explore research issues and some of the initial work for developing adaptive ESP.