Sensor Data Integration Using Ontologies for Event Detection.

Nowadays, computing applications operate in environments with multiple and heterogeneous data sources, such as data generated from IoT devices. Without contextual information, the information derived from these isolated data sources may cause bias, error, or a lack of correct comprehension. Data integration can help to promote a holistic view of data and support getting the most trustful meaning from the information. This work proposes an architecture in which ontologies help to provide context for data integration. Furthermore, ontologies and complex network concepts enrich context awareness and derive relations among data to identify events of interest. The approach is evaluated in a controlled experiment using real data from hydrological and hydrometric sensors. The results indicate it is possible to detect context and relate events from different data sources to new significant events through ontology and graph network analysis.