DIR workshop ontology stream data access

Date: 16/02/2011

Ontology-based Access to Streaming Data Sources

Jean-Paul Calbimonte

Ontology Engineering GroupFacultad de Informática, Universidad Politécnica de Madrid.

[email protected]

Data Intensive Research WorkshopEdinburgh, UK

mailto:[email protected];[email protected]

Introduction & Scope

2

• Streaming Data(t9, a1, a2, ... , an)(t8, a1, a2, ... , an)(t7, a1, a2, ... , an)......(t1, a1, a2, ... , an)......

Streaming Data

• Continuously appended data• Potentially infinite• Time-stamped tuples• Continuous queries• Changes of values over time• Latest used in queries

• Ubiquitous data capture• Data processing • Cheap• Noisy, Unreliable• Low computational, power resources, storage

• Sensor technologies

Enabling Ontology-based Access to Streaming Data Sources

• Applications in security surveillance, healthcare provision, environmental monitoring, you name it.

Motivation

3Enabling Ontology-based Access to Streaming Data Sources

Emergency planner

Flood risk alert: South East England

...

...

...

Real-time data

Meteorological forecasts

Flood defences data

Other sources

• Detect conditions likely to cause a flood• Present data model in terms of the user domain: e.g. Flood risk assessment

Example:• “provide me with the wind speed observations average over the last minute in

the Solent region, if it is higher than the average of the last 2 to 3 hours”

Wave,Wind,Tide

Motivation


• Ontologies can be used as such a common model• Achieve logical transparency in access to data.• Hide to the user where and how data are stored.• Present to the user a conceptual view of the data.• Use a semantically rich formalism for the conceptual

view.

[Calvanese2007]

• Need to provide solution for:• Establish mappings between ontological models and

streaming data source schemas• Access streaming data sources through queries over

ontology models

Background – Ontology-based Data Access


Ontology-based Data Access

R2O + ODEMapsterD2RQSquirrelRDFRDBToOntoRelational.OWLSPASQLVirtuosoMASTRO

Query/Data Transformation

Ontological Models

Relational Schemas

Mapping Document

Generate Semantic Web content from existing relational data sources

e.g. SPARQL

e.g. SQL

Linked DataRDF

Web of Data


Background – Querying Relational Data Streams


Streaming Data

STREAMAurora/BorealisCougarTinyDBSNEE

e1

e2e1

e2 e3

e2e1 e3

e1

e4

t t+1 t+2 t+3 t+4 t+5

WINDOW [tnow TO tnow-2] SLIDE 1

Transform infinite sequence of tuples to bounded bag

Window-to-Stream operators: convert stream of windows to stream of tuples

...

...

CQLSNEEqlTinyQL

Query engines

Query languages

Event Streams/Acquisitional Streams

Ontology-based Streaming Data Access


Querytranslation

Query Evaluator

Clie

nt Stream-to-Ontologymappings

SPARQLStream (Og)

[tuples]

Stream Engine (S3)

Ontology-based Streaming Data Access Service

RelationalDB (S2)

Sensor Network (S1)

RDF Store(Sm)

SPARQLStream algebra(S1 S2 Sm)

Datatranslation

q

[triples]

SNEEql



• Mappings from relational streams to ontological concepts• Extend stored data schema mappings• Study translation semantics

• Provide with a stream query language at ontological level• Use notion of RDF stream• Extend SPARQL• Window operator, window-to-stream operators

SPARQLStream


• RDF-Stream ......( <si-1,pi-1, oi-1>, ti-1 ),( <si, pi, oi>, ti ),( <si+1,pi+1, oi+1>, ti+1 ),......

Example:• “provide me with the wind speed observations over the last minute in the Solent Region ”

cd:Observation

xsd:double

cd:observationResult......( <ssg4e:Obs1,rdf:type, cd:Observation>, ti ),( <ssg4e:Obs1,cd:observationResult,”34.5”>, ti ),( <ssg4e:Obs2,rdf:type, cd:Observation>, ti+1 ),( <ssg4e:Obs2,cd:observationResult,”20.3”>, ti+1 ),......

STREAM <http://www.semsorgrid4env.eu/ccometeo.srdf>

SPARQLStream


Example:• “provide me with the wind speed observations over the last minute in the Solent Region ”

cd:Observation

xsd:double

cd:observationResult

PREFIX cd: <http://www.semsorgrid4env.eu/ontologies/CoastalDefences.owl#>PREFIX sb: <http://www.w3.org/2009/SSN-XG/Ontologies/SensorBasis.owl#> PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> SELECT ?windspeed ?windtsFROM STREAM <http://www.semsorgrid4env.eu/ccometeo.srdf> [ NOW – 1 MINUTE TO NOW MINUTES ] WHERE { ?WindObs a cd:Observation;

cd:observationResult ?windspeed; cd:observationResultTime ?windts;cd:observedProperty ?windProperty;cd:featureOfInterest ?windFeature.

?windFeature a cd:Feature;cd:locatedInRegion cd:SolentCCO.

?windProperty a cd:WindSpeed. }

cd:Feature

cd:featureOfInterest

cd:Property

cd:observedProperty

cd:locatedInRegion

cd:Region



Querytranslation

Query Evaluator

Clie


SPARQLSream (Og)

[tuples]

Stream Engine (S3)


RelationalDB (S2)

Sensor Network (S1)

RDF Store(Sm)

SPARQLSream algebra(S1 S2 Sm)

Datatranslation

q

[triples]

SNEEql

S2O Mappings


<conceptmap-def id="Observation_wind"name="http://www.semsorgrid4env.eu/ontologies/CoastalDefences.owl#Observation" virtualStream="http://www.semsorgrid4env/ccometeo.srdf"><uri-as>

<operation oper-id="concat"><arg-restriction on-param="string1">

<has-value>http://www.semsorgrid4env.eu/data#ObservationWind</has-value></arg-restriction><arg-restriction on-param="string2">

<has-column>meteostream.DateTime</has-column></arg-restriction>

</operation></uri-as>

<attributemap-def name="http://www.semsorgrid4env.eu/ontologies/CoastalDefences.owl#observationResult" dataType="xsd:double"><selector><aftertransform><operation oper-id="constant"><arg-restriction on-param="const-val"><has-column>meteostream.Hs</has-column>

</arg-restriction></operation>

</aftertransform></selector></attributemap-def>

cd:Observation

xsd:double


meteostream

Datetime: longHs : floatLon: floatLat: float

S2O Mappings


cd:Observation

xsd:double

cd:observationResult envdata_hornsea


envdata_westbay


envdata_milford


envdata_rhylflats


U

U

U

<union name="meteostream"><map value="http://www.semsorgrid4env.eu/ontologies/

CoastalDefences.owl#SolentCCO"><extent name="envdata_hornsea"></extent><extent name="envdata_rhylflats"></extent><extent name="envdata_milford"></extent><extent name="envdata_westbay"></extent>

</map> </union>

• Union of extents-streams



Querytranslation

Query Evaluator

Clie


SPARQLSream (Og)

[tuples]

Stream Engine (S3)


RelationalDB (S2)

Sensor Network (S1)

RDF Store(Sm)

SPARQLSream algebra

Datatranslation

q

[triples]

SNEEql

Query Translation


• Queries: SELECT ?y WHERE{ ?x a cd:Observation;

cd:observationResult ?y. }

SELECT ?yFROM STREAM < STREAM <http://www.semsorgrid4env.eu/ccometeo.srdf> [ NOW – 1 MINUTE TO NOW MINUTES ] >WHERE{ ?x a cd:Observation;

cd:observationResult ?y. }

Présentateur

Commentaires de présentation

q(x)\leftarrow Observation(x) \wedge observationResult(x,y)\\ q(\vec{x})\leftarrow \phi(\vec{x},\vec{y})) \mathit{\Psi \leadsto \Phi}

Query Translation


• Mappings: query algebra expression

over sources

cd:Observation

xsd:double


envdata_milford


envdata_milford

DatetimeHs

Présentateur

Commentaires de présentation

(Observation(x) \wedge \\observationResult(x,y))[t_i,t_f,\delta]\\

Query Translation


envdata_rhylflats

DateTimeHs

cd:Observation

xsd:double


envdata_hornsea


envdata_westbay


envdata_milford


envdata_rhylflats


U

U

U

envdata_milford

DateTimeHs

envdata_hornsea

DateTimeHs

envdata_westbay

DateTimeHs

DateTimeHs



Querytranslation

Query Evaluator

Clie


SPARQLSream (Og)

[tuples]

Stream Engine (S3)


RelationalDB (S2)

Sensor Network (S1)

RDF Store(Sm)

SPARQLSream algebra

Datatranslation

q

[triples]

SNEEql

vv

vv

Query Execution

20

PREFIX cd: <http://www.semsorgrid4env.eu/ontologies/CoastalDefences.owl#>PREFIX sb: <http://www.w3.org/2009/SSN-XG/Ontologies/SensorBasis.owl#> PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> SELECT ?waveheight ?wavets ?lat ?lonFROM STREAM <http://www.semsorgrid4env/ccometeo.srdf> WHERE { ?WaveObs a cd:Observation;

cd:observationResult ?waveheight; cd:observationResultTime ?wavets;cd:observationResultLatitude ?lat;cd:observationResultLongitude ?lon;cd:observedProperty ?waveProperty;cd:featureOfInterest ?waveFeature.

?waveFeature a cd:Feature;cd:locatedInRegion cd:SouthEastEnglandCCO.

?waveProperty a cd:WaveHeight. }

(SELECT Lon,timestamp,Hs,Lat FROM envdata_rhylflats) UNION (SELECT Lon,timestamp,Hs,Lat FROM envdata_hornsea) UNION (SELECT Lon,timestamp,Hs,Lat FROM envdata_milford) UNION (SELECT Lon,timestamp,Hs,Lat FROM envdata_chesil) UNION (SELECT Lon,timestamp,Hs,Lat FROM envdata_perranporth) UNION (SELECT Lon,timestamp,Hs,Lat FROM envdata_westbay) UNION (SELECT Lon,timestamp,Hs,Lat FROM envdata_pevenseybay)

envdata_rhylflats

Timestamp: longHs : floatLon: floatLat: float

envdata_hornsea

Observation

WaveHeightProperty

observedProperty

hasObservationResult

xsd:float

locatedInRegion

Ontologies Streams

S2O Mapping

envdata_milford

envdata_chesil

envdata_westbay

Region

Feature

SPARQLStreamSNEEql




Querytranslation

Query Evaluator

Clie


SPARQLSream (Og)

[tuples]

Stream Engine (S3)


RelationalDB (S2)

Sensor Network (S1)

RDF Store(Sm)

SPARQLSream algebra

Datatranslation

q

[triples]

SNEEql

Data Translation

• Data translation:Tagged tuples SPARQL bound variables

<ns9:sparql><ns9:head>

<ns9:variable name="waveheight"/><ns9:variable name="wavets"/>

</ns9:head><ns9:results>

<ns9:result><ns9:binding name="waveheight">

<ns9:literal datatype="http://www.w3.org/2001/XMLSchema#double">4.850</ns9:literal></ns9:binding><ns9:binding name="wavets">

<ns9:literal datatype="http://www.w3.org/2001/XMLSchema#long">1272588663</ns9:literal></ns9:binding>

</ns9:result><ns9:result><ns9:binding name="waveheight">

<ns9:literal datatype="http://www.w3.org/2001/XMLSchema#double">2.1230</ns9:literal></ns9:binding><ns9:binding name="wavets">

<ns9:literal datatype="http://www.w3.org/2001/XMLSchema#long">1272587400</ns9:literal></ns9:binding>

</ns9:result></ns9:results></ns9:sparql>

waveheight wavets4.850 12725886632.1230 1272587400

R2RML Mappings

• W3C RDB2RDF Working Group

• :waveObsResult rr:predicateMap [ rr:predicate ssg:observationResult ]; • rr:objectMap [ rr:column "Hs" ].• :waveObsTime rr:predicateMap [ rr:predicate ssg:observationResultTime ]; • rr:objectMap [ rr:column "timestamp" ].• :waveFoISea rr:predicateMap [ rr:predicate ssg:featureOfInterest ]; • rr:objectMap [ rr:object ssg:Sea ].• :waveObsProp rr:predicateMap [ rr:predicate ssg:observedProperty ]; • rr:objectMap [ rr:object ssg:WaveHeight ]. •• :MilfordWaveObservation a rr:TriplesMapClass;• rr:SQLQuery "";• rr:subjectMap [ rr:column "envdata_milford.DateTime"; rr:class ssg:Observation; rr:graph

ssg:ccometeo.srdf ];• rr:tableName "envdata_milford";• rr:predicateObjectMap :waveObsResult;• rr:predicateObjectMap :waveObsTime;• rr:predicateObjectMap :waveFoISea;• rr:predicateObjectMap :waveObsProp;


SemSorGrid4Env EU Project

1. Universidad Politécnica de Madrid, (UPM,

Spain)

2. University of Manchester (UNIMAN, UK)

3. National and Kapodistrian University of

Athens (NKUA, Greece)

4. University of Southampton (SOTON, UK)

5. Deimos Space SLU (DMS, Spain)

6. EMU Ltd. (EMU, UK)

7. TechIdeas (TI, Spain)

24

3

3

1


Project Challenges

• Integrated information space • Discovery new sensor networks• Integrate with existing ones • Integrate possibly other data sources

(e.g., historical databases)

020406080

100

1ertrim.

3ertrim.

EsteOesteNorte

s e n s o r n e tw o rk s

le g a c y d a ta s o u rc e s

s e m a n tic d a ta in te g ra tio n a n d q u e ry in g

th in a p p lic a tio n s (m a s h u p s )

re g is trie s

m id d le w a re

• Rapid development • flexible and user-centric decision

support systems • Use data from multiple

autonomous independently deployed sensor networks and other applications.


Implementation


• Design, implement and deploy a Semantic Integration Service

• Extend existing ontology-based data integration models to take into account sensor networks streaming data, semantic heterogeneity and quality of service

Implementation


Integration Query Service

IntegrationInterface

QueryInterface

PullInterface

Translator

ClientSNEE-WS

S2O mappingsrepository

SNEEql

IntegrateAs (DataResourceAddressList, S2O document)

DataResourceAddress

SPARQLQueryFactory (DataResourceName, SPARQLSTR )

DataResourceAddressaddQuery

GetStreamNewestItem(DataResourceName )

DataSetgetResultSetrepeat

CCO-WS

Implementation

• Flood Warning Application


- Wave Height updated every 10 min - Live data from sensor in buoys - Over 30 deployments

Conclusions

Ontology-base data access• Define stream extensions for R2O• Define SPARQLStream language syntax and semantics• Define S2O mappings-based translation semantics

Implementation• Enable engine support for « S2O » documents,

SPARQLStream queries• Enabled engine support for SNEEql translation and

connection• Limited to non-distributed scenario initially


30

Future Work

• Ontology-based data access• SPARQL construct expressions, aggregates, projected

operators• Implement adapters for other streaming sources• Add query rewriting algorithms

• Ontology-based streaming data integration• Horizontal & vertical integration• Integrate streaming + stored data• RDF data sources integration

• Streaming query optimization• Analyze cost models• Streaming sources statistics and metadata

• Quantitative evaluation



Planet Data

• EU FP7 Network of Excellence

• Working with EPFL Global Sensor Network

• SwissExperiment Data

Co-authors of this work

• Oscar Corcho – Universidad Politécnica de Madrid• Alasdair Gray – The University of Manchester

• Supported by the SemSorGrid4Env FP7-223913


Thanks!

33

Ontology-based Access toStreaming Data Sources

Jean-Paul [email protected]


Date post:	20-Jan-2015
Category:	Technology
Upload:	jean-paul-calbimonte
View:	824 times
Download:	1 times

DIR workshop ontology stream data access

Technology