PhD defense of Roland Stühmer
Roland Stühmer from PLAY has his PhD defense in Karlsruhe on July, 17th:
I am pleased to invite you to my PhD defence which will take place on Thursday July 17th at 1 p.m. in Karlsruhe at KIT.
Title: Web-oriented Event Processing
The jury is composed of the following people:
Prof. Dr. Rudi Studer, Karlsruher Institut für Technologie (KIT)
Prof. Dr. Opher Etzion, Academic College of Emek Yezreel
Prof. Dr. Thomas Setzer, Karlsruher Institut für Technologie (KIT)
Prof. Dr. Jan Kowalski, Karlsruher Institut für Technologie (KIT)
Abstract:
Event processing – computing performed on occurrences happening in a system or domain – is a common methodology of dealing with real-time data where situations must be detected in real-time. Event processing research and products today provide a good understanding and support for closed-domain systems such as enterprises. On the Web real-time results also gain interest as more and more data are available in data streams. Examples are social activity streams or sensor readings. Requirements for event processing on the Web are different from those in closed-domain systems. The question then arises as to howtheWeb can be made situationaware. In this thesis we collect the requirements for a Real-time Web and answer the posed question by contributing the design and realisation of aWeb-oriented event processing system to manage events, streams and queries.
The presented result is a semantic system that serves as an Event Marketplace: heterogeneous events from the Web modelled in RDF are matched and integrated using a processing language we designed and describe in this work. The system consists of these main components: an event processing layer to combine, integrate, filter and derive events quickly, in memory, and a storage layer to maintain historic events augmenting the real-time layer with long-term queries. Event-driven applications are implemented on this architecture by expressions in our language supporting hybrid queries combining both real-time queries (on pushed data) and historic queries (on pull data). A governance component enforces efficient access control on event streams and storage. The main contributions of the system design are its Web-orientation by adhering to open and extensible standards, its processing language offering the combination of real-time and historical queries on events and its governance capabilities creating a multi-tenant system based on permissions. The system usesWeb technologies such as Linked Data, RDF and SPARQL to model, organize, locate, process and control access to events.
We evaluate the artefacts produced as part of this work using qualitative and quantitative measures: Qualitative comparisons are made with the State of the Art and the overall cost of ourWeb-based approach is determined quantitatively and compared to a non-Web-based solution as the baseline.











