Generic Semantization of Vehicle Data Streams

To maximize the immediate use of the wide variety of vehicle data streams, we require to interpret their meaning and express it semantically. On the one hand, existing approaches to analyze sensor data are often use-case-specific and do not consider its streaming-nature. On the other hand, current semantic models for vehicle data focus on adding metadata to individual observations, missing out on representing the observations' actual meaning over time. Prioritizing only one of these two sides prevents the solution's generalization or creates dependencies on resources that are not available. This paper proposes an approach for a generic semantization of sensor data streams. The presented solution consists of methods to analyze and semantically annotate any sensor data stream of the type continuous or categorical. Compared to existing approaches, it enables a straightforward implementation of analytical queries at a higher abstraction layer and reduces, in most cases, the number of annotations per stream.