and Measurements Using Semantic Technologies Student: Alexandra - PowerPoint PPT Presentation

Enrichment of Sensor Descriptions and Measurements Using Semantic Technologies Student: Alexandra Moraru Mentor: Prof. Dr. Dunja Mladeni ć

Environmental Monitoring automation Traffic integration Monitoring Interoperability Building Monitoring Health Industrial Process Monitoring Monitoring 2 Images source: M. Botts, G. Percivall, C.Reed, J. Davidson, OGC SWE: Overview And High Level Architecture

Motivation • Understanding and managing of sensor data – We want a way of doing sensing that can make the data available to any application that needs that specific data [1] – Challenges: • associate meaning to sensor data • computer understandable representation • many communities participate in sensor deployments • Semantic Technologies – identified as key enabling technologies for sensor networks (W3C) – semantic enrichment can be considered as a first step [1] John Cox, Turning the world into a sensor network, Network World, August 11, 2010. 3

Contributions • Semantic enrichment of sensor descriptions and measurements – Definition of a framework – Instantiation of framework components – Examples of applications 4

Part I Framework Definition • Introduction • Problem Description • Framework Components 5

Introduction (1/2) • Sensor – material or device which changes its properties according to a physical stimulus – can be attached to more complex devices – sensor nodes • computing and communication capabilities • embedded into physical objects • wired and wireless networks 6

Introduction (2/2) • Internet of Things – world-wide network of heterogeneous smart objects – sensors, actuators, RFIDs, MEMS – based on standard communication protocols – focused on establishing connectivity • Web of Things – integrating smart objects into the Web – a.k.a Sensor Web, Physical Web – based on standards like HTML, XML, RSS – focused on application layer 7

Problem Description (1/2) • Integration of sensor data from different systems • Provide machine understandable representation of data – Describe the meaning of data and the context in which it was collected • Environment characteristics • Sensor properties 8

Problem Description (2/2) • Apply semantic technologies to sensor web – Enriching the sensor data • enrichment of data generally refers to adding information • semantic enrichment refers to associating semantic tags – Publishing annotated sensor data • enables the development of new applications • through standardized web services, application specific methods – requires prior knowledge of the infrastructure used • following Linked Open Data (LOD) principles 9

Conceptual Framework • Framework for semantic enrichment of sensor data – automatizing the process enriching sensor descriptions and measurements Ontology Extension Query End-Point Ontology Semantic Browsers Descriptions Collection Enrichment Semantic Measurements Repository preprocess and Inference Engines of Sensor enrichment Data Sensor Descriptions Enrichment Data and Measurements Components Consumers 10

Framework Components (1/3) • Sensor Descriptions and Measurements – Sensor description refer to the metadata defining sensor characteristics – Sensor measurements – numerical Ontology Extension values quantifying the changes of sensor properties Ontology Collection • Ontology Collection – set of ontologies necessary for describing sensor characteristics and providing context for sensor measurements. Sensor Descriptions and Measurements 11

Framework Components (2/3) • Enrichment Components – sensor descriptions are enriched with semantic concepts – sensor measurements are processed to generate new features which are then enriched by semantics. Ontology Extension • Steps of the enrichment process – Analysis of the sensor descriptions Ontology Descriptions and measurements Collection Enrichment – Selection of ontologies – Extension of the selected ontologies Measurements preprocess and with concepts specific to the domain of enrichment application – Implementation of enrichment Sensor Descriptions Enrichment and Measurements components Components 12

Framework Components (3/3) • Data Consumers • Semantic Repository of Sensor Data – query end-points – semantic browsers – contains the enriched sensor descriptions and measurements. – Inference engines Ontology Extension Query End-Point Ontology Semantic Browsers Descriptions Collection Enrichment Semantic Measurements Repository preprocess and Inference Engines of Sensor enrichment Data Sensor Descriptions Enrichment Data and Measurements Components Consumers 13

Part II Instantiation of Framework Components • Sources of Sensor Data – Non-standardized – Standardized • Ontology Collection – OWL ontologies – Cyc ontology • Architecture • Enrichment Components • Semantic Repository of Sensor Data 14

Sources of Sensor Data (1/2) • Non-standardized dataset – data collected from a sensor network for monitoring environmental conditions • temperature, humidity, luminance and pressure – centralized MySQL database server, • both the meta-data and sensor measurements . 15

Sources of Sensor Data (2/2) • Standardized dataset – contains description and measurements of sensors in the area of ocean tides and currents • air temperature, water temperature, water level, currents, wind, air pressure, salinity • sensor description (SensorML) • sensor measurements (O&M) – 751 sensor nodes, 1379 sensors measuring 14 types of properties – downloaded and processed offline 16

Ontology Collection (1/3) • OWL ontologies – W3C Semantic Sensor Network ontology (infrastructure) 17

Ontology Collection (2/3) • Additional (external) OWL ontologies – Basic GeoWGS84 Vocabulary, provides namespaces for representing coordinates – Geonames, provides geographical names in RDF representation • findNearbyPlacename web service – W3C time ontology • defining time intervals for sensor measurements 18

Ontology Collection (3/3) • Research Cyc 19

Architecture SPARQL Endpoint SESAME OWL Ontologies SSN Ontology Jena Framework Pubby Data Publishing RDF descriptions of sensors MySql JDBC Database JAXB SensorML API descriptios ResearchCyc OntoGenUI OntoGenUI OntoGen O&M measurements 20

Enrichment Components (1/4) • Enrichment of sensor data – Input: sensor data + ontologies – manually creating rules • extract the information from the datasets • attach the corresponding semantic concepts – for the non-standardized dataset -> OWL ontologies – for the standardized dataset -> both collections of ontologies, separately – Output: RDF representation of the original dataset • annotated with semantic concepts from the ontology collection 21

Enrichment Components (2/4) • Non-standardized dataset – Database 22

Enrichment Components (3/4) • Standardized dataset – platforms described in SensorML -> instances of Platform . – the networks to which these platforms belong to -> instances of Deployment . – the platforms components -> instances of SensingDevice . – the observed properties of sensing devices -> instances of the subclasses extending Property • related to the sensed domain by the using the relation isPropertyOf and the subclasses extending FeatureOfInterest . – the geographical locations of the platforms, given by latitude and longitude coordinates -> lat and long relations from the GeoWGS84 vocabulary. • findNearbyPlaceName web service (fro GeoNames) -> finds the name of the closest populated place to the platform location – Computation based enrichment of measurements 23

Enrichment Components (4/4) • Standardized dataset – Enrichment of measurements – data mining tool for processing the sensor measurements and for extracting knowledge from the raw measurements – enriched measurements annotated according to the collection of OWL ontologies • exported in RDF format – permits the user to take advantage of knowledge extracted from the raw measurements • Features generated from the raw sensor measurements – wind and sea conditions for sailors according to the Beaufort scale • 26 nominal values, such as: Calm, Flat, Fresh Breeze , etc. – migraines caused by atmospheric pressure according to pressure values published in medical studies • risk of headache: NoHeadache, Headache and HighHeadache • time of day intervals: early morning, late evening , etc. 24

Semantic Repository of Sensor Data • Implementation of Semantic Repository of Sensor Data – Sesame, framework for processing RDF data • store, parse, query, perform inference on RDF data • Java API (storing and updating the enriched sensor data) • for each dataset a separate repository 25

Part III Applications • Sensor search • Data Publishing 26

Sensor Search (1/3) • Finding specific sensors, from which one could be interested in gathering data • Searching directly through sensor measurements for explicit values or different events. • Formulating queries for retrieving the results we need – SPARQL end-points 27