SEPP: Semantics-Based Management of Fast Data Streams

Abstract

In the era of big data processing there is an emerging need for methodologies supporting the management of data-intensive application scenarios. Complex Event Processing is an integral part of many fast data application as an underlying technology for event correlation and pattern detection. Increased volume of event streams as well as the demand for more complex real-time analytics require for execution of processing pipelines among heterogeneous event processing engines. In this paper, we propose a semantic model for the management of fast data streams using the concept of Semantic Event Processing Pipelines (SEPP). We provide methodology, architecture and language for semantic discovery and binding of real-time processing services from arbitrary stream processing engines. Our approach aims to improve reusability of real-time processing services by providing high-level interfaces to stream processing implementations. By these means this work paves the way for an easier development and management of real-time big data applications.