It’s a Machine Learning world: no future for Complex Event Processing?

Abstract

In 2010, Complex Event Processing was a groundbreaking approach for data processing for Early Warning Systems and Decision Support Systems in the crisis management domain. Although Machine Learning has eclipsed CEP in the ISCRAM community these last few years, these two approaches do not fulfil the same objectives. With this article, a quantitative analysis is made on papers related to these domains to understand the current trends and identify the next challenges the ISCRAM community should address. Combining CEP with ML can help to overcome the strong assumptions made towards the CEP rules this last decade, such as the availability of experts, structured data, etc. It is then recommended that the ISCRAM community explore the hybridisation of these approaches to improve the responsiveness in real time of EWS and DSS in crisis situations, emphasising the automatic, non-supervised generation of CEP rules.