Dagstuhl

2. Outline
™ Motivation
™ The Model stRDF
™ The Query Language stSPARQL
™ The Scalable Geospatial RDF Store Strabon
™ Experimental Evaluation
™ Other Extensions to RDF
™ Conclusions

3. Motivation
™ Need for modeling of
™ Geospatial and temporal information
™ Product metadata
™ Product content
™ Need to link to other data sources
™ GIS data
™ Other information on the Web

4. TELEIOS Project
™ TELEIOS EU project
™ Development of a Virtual Earth Observatory
™ Accessible to expert and non-expert users
™ Scalable (Storing/Querying PBs)
™ Use Cases
™ DLR: A Virtual Observatory for TerraSAR-X data
™ NOA: Real-time fire monitoring based on continuous
acquisitions of EO images and geospatial data
™ Technologies
™ NKUA: Geospatial extensions for RDF and SPARQL
™ CWI: Array extensions to SQL

5. The Data Model stRDF
™ stRDF extends RDF with:
™ Spatial literals encoded by Boolean combinations of
linear constraints
™ New Datatype for spatial literals (strdf:geometry)
™ Valid time of triples encoded by Boolean combinations
of temporal constraints
™ stRDF (most recent version, TELEIOS)
™ Spatial literals encoded in Well-Known Text/GML
(OGC standards)
™ Valid time of triples ignored for the time being

6. stRDF: An example

7. stRDF: An example
ex:BurntArea1 rdf:type noa:BurntArea.
ex:BurntArea1 noa:hasID "1"^^xsd:decimal.
ex:BurntArea1 noa:hasArea "23.7636"^^xsd:double.
ex:BurntArea1 strdf:geometry "POLYGON(( 38.16 23.7, 38.18
23.7, 38.18 23.8, 38.16 23.8, 38.16 23.7));<http://
spatialreference.org/ref/epsg/4121/>"^^strdf:WKT .
Spatial Data Type
Well-Known Text
Spatial Literal

8. stRDF: An example (GML)
ex:BurntArea1 rdf:type noa:BurntArea.
ex:BurntArea1 noa:hasID "1"^^xsd:decimal.
ex:BurntArea1 noa:hasArea "23.7636"^^xsd:double.
ex:BurntArea1 strdf:geometry "<gml:Polygon
srsName='http://www.opengis.net/def/crs/EPSG/0/4121'>
<gml:outerBoundaryIs><gml:LinearRing><gml:coordinates>38.16
,23.70 38.18,23.70 38.18,23.80 38.16,23.80
38.16,23.70</gml:coordinates></gml:LinearRing></
gml:outerBoundaryIs></gml:Polygon>"^^strdf:GML .
Spatial Data Type
GML
Spatial Literal

9. stSPARQL: An example
™ Find all burnt forests within 10km of a city
SELECT ?BA ?BAGEO
WHERE {
?R rdf:type noa:Region .
?R strdf:geometry ?RGEO .
?R noa:hasCorineLandCoverUse ?F .
?F rdfs:subClassOf clc:Forests .
?CITY rdf:type dbpedia:City .
?CITY strdf:geometry ?CGEO .
?BA rdf:type noa:BurntArea .
?BA strdf:geometry ?BAGEO .
FILTER(strdf:Intersect(?RGEO,?BAGEO) &&
strdf:Distance(?BAGEO,?CGEO) < 0.02) }
Spatial
Functions

10. stSPARQL: Geospatial SPARQL 1.1
™ We define a SPARQL extension function for each function
defined in the OpenGIS Simple Features Access standard
™ Basic functions
™ Get a property of a geometry (e.g., strdf:srid)
™ Get the desired representation of a geometry (e.g., strdf:AsText)
™ Test whether a certain condition holds (e.g., strdf:IsEmpty,
strdf:IsSimple)
™ Functions for testing topological spatial relationships
(e.g., strdf:equals, strdf:intersects)
™ Spatial analysis functions
™ Construct new geometric objects from existing geometric objects
(e.g., strdf:buffer, strdf:intersection, strdf:convexHull)
™ Spatial metric functions (e.g., strdf:distance, strdf:area)
™ We define spatial aggregate functions (e.g., strdf:union,
strdf:extent)

11. stSPARQL: Geospatial SPARQL 1.1
™ Spatial terms
™ Constants (e.g., "POLYGON((38.16 23.7, ...)) " ^^strdf:WKT)
™ Variables (e.g., ?GEO)
™ Results of set operations (e.g., strdf:intersection, strdf:union,)
™ Results of geometric operations (e.g., strdf:boundary, strdf:buffer)
™ Select clause
™ Construction of new geometries (e.g., strdf:buffer(?geo, 0.1))
™ Spatial aggregate functions (e.g., strdf:extent(?geo))
™ Metric functions (e.g., strdf:area(?geo))
™ Filter clause
™ Functions for testing topological spatial relationships between spatial terms (e.g., strdf:contains(?
G1, strdf:union(?G2, ?G3)))
™ Numeric expressions involving spatial metric functions (e.g., strdf:area(?G1) ≤ 2*strdf:area(?G2)+1)
™ Boolean combinations
™ Having clause
™ Boolean expressions involving spatial aggregate functions and spatial metric functions or functions
testing for topological relationships between spatial terms (e.g., strdf:area(strdf:union(?geo) ) > 1 )
™ Updates
™ Similar to the OGC standard GeoSPARQL

12. Strabon: A Scalable Geospatial
RDF Store
™ Storing: stRDF
™ Querying: stSPARQL, GeoSPARQL
™ Storage Backend: PostGIS, MonetDB
™ Extends: Sesame 2.6.3
™ Open Source: http://strabon.di.uoa.gr/
™ EU Projects: Semsorgrid4Env and TELEIOS

13. Strabon: A Scalable Geospatial
RDF Store
stRDF
graphs
stSPARQL/
GeoSPARQL
queries
WKT GML
Query Engine
Parser
Optimizer
Evaluator
Transaction
Manager
Storage Manager
Repository
SAIL
RDBMS
Strabon
PostGIS
GeneralDB

14. Query Evaluation
Sesame’s Native Store (Baseline)
™ Index (Geometries): None
(geometries treated as ordinary literals)
™ Query Evaluation
1. Triple pattern evaluation
2. Evaluation of spatial functions in Sesame using JTS library
Strabon + PostGIS
™ Index (Geometries): R-Tree
™ Query Evaluation
PostGIS optimizer
Evaluation of spatial functions in PostGIS
Triple pattern evaluation might come before/after evaluation of spatial
functions

15. Experimental Evaluation
™ Workload based on geospatial linked datasets
q 150 million triples
q Test queries based on logs from Dbpedia,
LinkedGeoData
1. Detailed evaluation is coming up
[ISWC paper submission under preparation]

16. Datasets: Characteristics
DBpedia GeoNames LGD Pachube SwissEx CLC GADM
Size 7.1GB 2.1GB 6.6GB 828KB 33MB 14GB 146MB
Triples 58,722,893 17,688,602 46,296,978 6,333 277,919 19,711,926 255
Spatial
Terms
386,205 1,262,356 5,414,032 101 687 2,190,214 51
Distinct
Spatial
Terms
375,087 1,099,964 5,035,981 70 623 2,190,214 51
Points 375,087 1,099,964 3,205,015 70 623 - -
Linestrings - - 353,714 - - - -
Polygons - - 1,704,650 - - 2,190,214 51
(79,831 vertices)

17. LinkedGeoData
™ Q1: Find the names of the points of interest located in the
city of Rotterdam.
SELECT ?poiName
WHERE {
?poi rdf:type lgd:Amenity .
?poi rdfs:label ?poiName .
?poi geo:geometry ?poiGeo .
FILTER (strdf:inside(?poiGeo, "POLYGON((
3.9111328125 51.7401123046875, 3.9111328125 52.05322265625,
4.6087646484375 52.05322265625, 4.6087646484375 51.7401123046875,
3.9111328125 51.7401123046875))"^^strdf:WKT)) }

18. LinkedGeoData/GADM
™ Q2: Find the points of interest located in Luxembourg.
SELECT ?label
WHERE {
?poi rdf:type lgd:Amenity .
?poi rdfs:label ?label .
?poi geo:geometry ?poiGeo .
gadm:LUXEMBOURG gadm:hasGeometry ?luxGeo .
FILTER (strdf:inside(?luxGeo, ?poiGeo)) }

19. Response Time for Queries Q1, Q2
0
50
100
150
200
250
cold caches warm caches cold caches warm caches
Q1 Q2
ResponseTime(seconds)
Workload based on Geospatial Linked Datasets
Strabon
Sesame

20. Experimental Evaluation
™ Workload based on a synthetic dataset
q Dataset based on OpenStreetMaps
q 10 million triples (2 GB) up to 1 billion triples (110 GB)
q Triples with spatial literals: 1.4 and 140 million triples
™ Response time of queries with various thematic and
spatial selectivities

21. Sample Data
“14”
“POINT(2.00 3.00)”
osm:id
geordf:hasGeometry1
osm:Node
rdf:type
osm:Tag osm:Tag1
“1”
osm:Node23
rdf:type
“1”
osm:key
osm:value
osm:hasTag

22. Sample stSPARQL Query
SELECT *
WHERE {
?tag geordf:key "1" .
?node geordf:hasTag ?tag .
?node geo:hasGeography1 ?geo .
FILTER (strdf:inside(?geo,
"POLYGON((-1 -1, 0.056568542 -1,
0.056568542 0.056568542, -1 0.056568542, -1 -1
))"^^http://strdf.di.uoa.gr/ontology#WKT
))
}

23. Experimental Results: 10 million
0.1
1
10
100
101
102
103
104
105
106
responsetime[sec]
number of Nodes in query region
Response Time (cold caches)
(postgres) key = 1
(postgres) key = 1024
(sesame) key = 1
(sesame) key = 1024

24. Experimental Results: 500 million
1
10
100
1000
10000
10
3
10
4
10
5
10
6
10
7
responsetime[sec]
number of Nodes in query region
Response Time (cold caches)
(postgres) key = 1
(postgres) key = 1024
(sesame) key = 1
(sesame) key = 1024

25. Other Extensions to RDF

26. The Framework RDFi
™ Extension of RDF with incomplete information
™ New kind of literals (e-literals) for each datatype
™ Property values that exist but are unknown or partially known
™ Partial knowledge: captured by constraints
(appropriate constraint language L)
™ RDF graphs extended to RDFi databases: pair (G, φ)
G: RDF graph with e-literals
φ: quantifier-free formula of L
™ Formal semantics for RDFi and SPARQL query evaluation
™ Representation System: CONSTRUCT with AUF graph patterns
™ Certain Answer: semantics, algorithms, computational complexity when L is
a language of spatial topological constraints
™ Implementation in the context of Strabon has started with L being PCL
(topological constraints between variables and polygon constants)
[conference submission]

27. Challenges
™ Query processing in face of linked geospatial
data
™ Query optimization for spatial data
™ Federation
™ Efficient query processing with incomplete
information

28. Conclusions
™ stRDF/stSPARQL
™ Model and query language for the representation and
querying of geospatial data
™ Strabon
™ Scalable Geospatial RDF Store
™ RDFi
™ Framework for representation and querying of RDF
data with incomplete information

29. Thanks
™ Strabon
http://strabon.di.uoa.gr/
™ NOA Application Demo
http://test.strabon.di.uoa.gr/NOA/
™ TELEIOS EU Project
http://www.earthobservatory.eu/