Published
NCIT2019
A Review on Complex Event Processing Approaches: Research Background and Challenges
Abstract
With rapidly increasing adoption of loT technologies in many business domains, data is flowing into distributed storages both on premise and off premise at an incredible rate and volume. To gain insight from these data in order to make a timely decision is not a trivial task, and a highly efficient and data complexity management are needed to enable such task. Traditional batch data processing focusing on calculation and processing on collection of data is not suitable for the task.
A data processing technique that can handle large and rapid data flow to enable real-time decision making must be considered instead. In this paper, we conduct a literature review and analyze the Complex Event Processing (CEP) approach which is the framework for continuous and real-time data processing. To this end, the continuous and streaming data processing approaches including (1) Information Flow Processing (IFP), (2) CEP advancements and use cases, (3) CEP and Machine Learning, and (4) Online Machine Learning are presented in the analytical way.
According to our comparative analysis, we found that Machine Learning (ML) has a potential to increase the efficiency in CEP. Specifically, Online ML is the method for continuous data processing with the capability to flexibly adjust the process rules. In addition, we found that unsupervised learning is a promising approach to support unlabeled data processing whereas the supervised leaming model requires Iabeled data in order to define the output. Finally, we give the concluding remarks on the reviewed approaches and point out the future works that are worth for further investigation.
A Review on Complex Event Processing Approaches: Research Background and Challenges. กรุงเทพ, NCIT2019 (169-175).