Association Rule Mining with Context Ontologies: An Application to Mobile Sensing of Water Quality

Internet of Things (IoT) applications by means of wireless sensor networks (WSN) produce large amounts of raw data. These data might formally be defined by following a semantic IoT model that covers data, meta-data, as well as their relations, or might simply be stored in a database without any formal specification. In both cases, using association rules as a data mining technique may result into inferring interesting relations between data and/or metadata. In this paper we argue that the context has not been used extensively for added value to the mining process. Therefore, we propose a different approach when it comes to association rule mining by enriching it with a context-aware ontology. The approach is demonstrated by hand of an application to WSNs for water quality monitoring. Initially, new ontology, its concepts and relationships are introduced to model water quality monitoring through mobile sensors. Consequently, the ontology is populated with quality data generated by sensors, and enriched afterwards with context. Finally, the evaluation results of our approach of including context ontology in the mining process are promising: new association rules have been derived, providing thus new knowledge not inferable when applying association rule mining simply over raw data.