Semantic annotation of summarized sensor data stream for effective query processing

Abstract

In the big data era, the volume of streaming data produced by sensor networks is staggeringly large that enables business intelligence to make well-informed decisions on emerging modern applications. Performing the data analytics and query processing over the fast arriving data streams is a tedious process. The semantic annotation of the data stream provides a high-level description, and a semantic context supports intelligent querying and data analytics. This paper presents a framework called SEmantic Annotation over Summarized sensOr Data stReam (SEASOR) that includes summarization, semantic annotation, and query processing that facilitates sensor data stream analytics. The summarization merges these types of stream values to increase the query performance and decrease the memory space. The semantic annotation is scripted with the help of application-dependent base ontology that extends the Semantic Sensor Network (SSN) ontology. The annotation of the sensor stream provides detailed descriptions for the observation of sensors using the base ontology, and it divides the streaming sensor data into several subsets according to the sensing features. The domain model enables the query processor to access the relevant results via an annotated Resource Description Framework (RDF). The query processor uses the extended SPARQL (Cs-SPARQL) to access only the relatively small subset via an annotated RDF file and allows extending the query processing to support windows and the parallel processing of data streams. The experimental results prove that the proposed SEASOR provides timely answers to the user queries and achieves better performance in terms of result accuracy by 95%.