A context-aware middleware for real-time semantic enrichment of distributed multimedia metadata

摘要

This paper investigates the problem of the real-time integration and processing of multimedia metadata collected by a distributed sensor network. The discussed practical problem is the efficiency of the technologies used in creating a Knowledge Base in real-time. Specifically, an approach is proposed for the realtime, rule-based semantic enrichment of lower level context features with higher-level semantics. The distinguishing characteristic is the provision of an intelligent middleware-based architecture on which low level components such as sensors, feature extraction algorithms, data sources, and high level components such as application-specific ontologies can be plugged. Throughout the paper, Priamos, a middleware architecture based on Semantic Web technologies is presented, together with a stress-test of the system's operation under two test case scenarios: A smart security surveillance application and a smart meeting room application. Performance measurements are conducted and corresponding results are exposed.