Stream Reasoning for Situation Detection

During the past decade, the Semantic Web has emerged as an evolution of the human readable web, towards a web of macine processable data with formal semantics. Currently, the trend is also towards processing of huge amounts of data, and highly dynamic data, e.g. sensor data.

Stream reasoning denotes techniques to do automated reasoning on an incoming stream of data fragments, in combination with existing background knowledge. Recently several approaches and languages for stream reasoning on RDF/OWL data, in a Semantic Web context, have also been proposed.

Situation detection, or sometimes called Complex Event Processing (CEP), is the process of detecting and reacting to high-level events, based on low-level input data, such as sensor readings. In the security domain, CEP is frequently used to detect ongoing violations or dangerous situations, however so far RDF/OWL models have not been used as the basis for this. Additionally, the notion of Ontology Design Patterns (ODPs) provide a means to create small and general RDF/OWL models that suit a specific task, but so far CEP is not supported by any such ODPs.

The aim of this master thesis is to analyze existing methods, languages and tools for RDF/OWL stream reasoning and CEP, and analyze their suitability and limitations for CEP in the security domain. Depending on the analysis results the thesis results could either include a proof-of concept implementation showing how existing methods can be used in this new domain, or some extensions/modifications of the existing methods/languages/tools could be proposed and implemented. An additional result should be an ODP (or a set of ODPs) supporting the general task of CEP in OWL.

Recommended skills to attempt this thesis include good knowledge of logical languages (preferably previous knowledge of RDF/OWL and/or Description Logics), web languages such as XML, and programming experience, preferably in Java.