Author(s): Aitor Corchero; Gabriel Anzaldi; Roberto García
Linked Author(s):
Keywords: Sensor Data; Semantic Web; Decision Support; Exploration; Stream Reasoning
Abstract: To date, Semantic Sensor Web research and development has focused on establishing common techniques and practices that homogenize how to discover, collect and integrate information from sensors. However, as solutions for this part of the problem are starting to become successful, huge databases of sensor data begin to accumulate. Therefore, the focus should now change to improve data management and reduce information overload, helping users discard irrelevant information and make its exploration easy and intuitive. This paper depicts the development of an architecture that focuses on water sensors data fusion, plus additional data from heterogeneous sources that contextualizes sensors output. By mixing and semantically annotating data, the architecture better assist users to autonomously understand data and then, generate new management strategies. The proposal is based on applying Semantic Streams Technology (SST) to collect and integrate continuous sensor data from multiple databases in form of RDF streams. The RDF streams are generated by transforming SQL-structured data into semantic data by using a knowledge base, thus enriching the information contained in the database. Hence, abstracting sensor data stored in a database through the R2RML technique generates the RDF streams. Moreover, SST processes sensor data using real-time semantic stream reasoning over the generated RDF streams from noisy sensor data to support the decision process even in large and complex scenarios. The proposed architecture has been analyzed by combining it with the infrastructure deployed over the water supply and distribution chain implemented in the context of the FP7-Wat ERP project, where sensor data is continuously stored in OGC-SOS server database. This data is transformed into RDF streams using also the Water Management Ontology developed in Wat ERP. The semantic stream reasoning is performed by C-SPARQL permitting the architecture to register and execute queries, which provide water managers knowledge to improve water resource management. Consequently, the proposed architecture enhances current state of the art in: (i) avoiding information replication by wrapping data streams with semantic stream technologies; and (ii) proving real-time reasoning, decision support and rapid actuation against resource mismanagements or other critical situations.
Year: 2015