Exploring the Use of the Semantic Web for discovering, retrieving and processing data from Sensor Observation Services

Exploring the Use of the Semantic Web for discovering, retrieving and processing data from Sensor Observation Services

De Liefde, I.

De Vries, M.E. (mentor)
Meijers, B.M. (mentor)

Architecture and The Built Environment

OTB

Geomatics

2016-06-28

Developments such as smart cities, the Internet of Things (IoT) and the Infrastructure for Spatial Information in Europe (INSPIRE) are causing a growing amount of observation data to be produced. The Open Geospatial Consortium (OGC) has developed Sensor Web Enablement (SWE) standards for modelling and publishing this data online. However, their use is currently limited to geo information specialists, who have knowledge about which data services are available and how to access them. With the use of the semantic web, online processes can automatically find and understand observation metadata. This opens up the SWE services to a large user audience. Therefore, this thesis has designed a conceptual system architecture that uses the semantic web to improve sensor data discovery as well as the integration and aggregation of sensor data from multiple sources. A conceptual system architecture is presented containing two web processes. The first process automatically creates an online semantic knowledge base of sensor metadata, by harvesting Sensor Observation Services. Metadata based on the Observations and Measurements (O&M) data model are retrieved, which includes the location of deployed sensors, what they observe, how they observe it and at which SWE service their data can be requested. The second process automatically translates logical queries of users into observation data requests. It also performs further processing before returning the observation data to the user. Both of these processes have been tested in a proof of concept implementation. A Web Processing Service (WPS) has been created as a proof of concept, making the two processes available online. The proof of concept is able to harvest sensor metadata, convert it to linked data, and publish it on the semantic web with links to and from other metadata. It results in a semantic knowledge base of sensor metadata, which improves the discovery in a machine understandable way. It allows multiple Sensor Observation Services to be used in combination with each other, due to the harmonisation involved in creating linked data. The WPS is therefore able to retrieve and process observation data from multiple Sensor Observation Services, using queries such as: What are the average particulate matter levels per month in neighbourhoods of Delft over the last five years? In conclusion, the presented conceptual system architecture contains two processes which improve the discovery, retrieval and aggregation of sensor data from multiple Sensor Observation Services using the semantic web. However, creating a completely automated process for harvesting metadata, which works with every Sensor Observation Service is not yet feasible, as non-meaningful identifiers are allowed to be used according to the SWE standards. Linked data ontologies should also be extended to include all available metadata. Currently there are no ontologies able to define the concept of a geo web service. Another conclusion is that the SPARQL endpoints used in this thesis cannot cope with complex vector geometries. This is due to their inability to handle verbose queries. Nevertheless, the proof of concept shows that the semantic web can enable current SWE services to be used in new ways, making observation data available to a larger user audience.

sensor web
semantic web
sensor web enablement
linked data
observation data

http://resolver.tudelft.nl/uuid:44d66025-3e51-4d32-81aa-e093f0aca833

Student theses

master thesis

(c) 2016 De Liefde, I.