Indexing and graphical analysis in a complex event processing environment

Abstract

In recent decades traditional Data Warehouses (DWHs) have come a long way. Triggered by the need of supporting strategic decision making, research has maintained a constant interest in developing and enhancing tools aimed at in depth analysis of historic data. In the meanwhile business itself has undergone major changes and new approaches are developed to adapt to these: Instead of monitoring and analyzing whole business structures, single processes have been moved into the spotlight, enabling operational workforce to come to well informed decisions. A privilege which is traditionally reserved to strategic management. As DWHs have problems in supporting the analysis of near real time data new concepts are sought for. As nowadays near-real-time processing is often done using complex event processing (CEP), visualization tools and analysis software is recently being developed in that context. Even though CEP is well established and has yielded a variety of highly sophisticated solutions, both scientific and industrial, the graphical access to the generated knowledge is still a fairly young field. This thesis discusses the need for (near-) real time analysis of ongoing business processes, as well as the possible solution CEP poses. The main attention is aimed at the need for sophisticated graphical analysis tools, enabling one sight understanding of underlying processes. The framework EventXm is presented, which poses a viable open source solution by combining text based query tools with interactive, graphical event stream mining and display functionality. The final section will discuss earlier presumptions, which announced CEP solutions to be future "DWH replacements", pointing out the benefits a symbiosis of both concepts may bring along.<br />