Robert Laurini on "Geographic Ontologies and Spatial Relations"

1. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
Example of ontology
Physical city
Aggregation or
Tesselation Network
Spatial
City-blocks relation Streets
Ontologies for geographic
1 – Introduction
applications
• 1 – Introduction • Oντος = being ; Λογια = discourse
• 2 – Theoretical bases of spatial ontologies • Aristotle: « The study of existing objects »
• 3 – Spatial relationships • Def1: theory of objects and their relations
• 4 – GeoOWL • Def2: theory of entities, especially of entities
• 5 – Gazetteers which exist in a language
• 6 – Conclusions • Def3: explicit specification of
conceptualization (Gruber)
INPUT 2012 1

2. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
Guarino’s definition Ontological commitment
• Nicola Guarino : “An ontology is • Usually, several definitions for the same
generally regarded as a designed artifact entity, for instance a horse, a table, etc.
consisting of a specific shared vocabulary •
used to describe entities in some domain • Several agents/shareholders agree to a
of interest, as well as a set of assumptions common definition of an object
about the intended meaning of the terms • Consensus about a definition
in the vocabulary” • Shared vocabulary
Concepts Example about roads
• Distinguish between terms and concepts
• Distance (km or mile) syntactic
• At mathematical level :
• Street or motorways semantic
Ontology = graph between concepts
= semantic network
INPUT 2012 2

3. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
We do have the road file! We do have the road file!
The road
The road file? The road Water Supply
file? Garbage men Postmen
Yes, I’ve it file? Company
Yes, I’ve it Yes, we’ve it
Private roads No Yes ??
Generally
Public roads Yes Yes
yes
Road with
? ? Yes
water supply
Road without
? ? No
Water supply
Total 234 251 241
Beginning of an urban ontoloy Beginning of an urban ontoloy
City Houses Apartments
Land space
Network Habitation Working places
Built space Non-built space Demography Resides Works
Human being
Building etc.
etc.
Spends Recreation
Shops Recreation places
INPUT 2012 3

4. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
2 – Theoretical bases of spatial
Examples about coast
ontologies
• Generic concepts
• Geographic objects
• Spatial relations
• Modeling
Courtesy Jonathan Raper of City University London, GISci 2002 Keynote
About Geographic Ontologies Geographic features
• Two definitions • Geographic features
– Conventional ontologies of geographic features – Crisp boundaries
(with is-a and part-whole relations) – Fuzzy boundaries
– Continuous fields
– Ontologies with spatial relationships between • Modeling
geographic features – Point, line, area volume
– Multi-representations
– Multi-scale
INPUT 2012 4

5. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
CONOSCENZE SPAZIALI E CONOSCENZE GEOGRAFICHE
Geographic Feature Types Fuzzy Geographic Objects
Membership grades (0 – 100 %)
• Geodetic objects
River: Major bed, major bed
• Administrative objects
• Human-made features
• Natural features
100%
– With known boundaries (crisp)
– With unknown boudaries (fuzzy) 80%
– With « no » boundaries (continuous fields) 60%
40%
R. Laurini
Geodetic Objects
Continent
• Theoretical objects onto the geoid
– Equator
Sea
– North and South Poles
– Meridians
– Parallels
Mangrove
• Modeled by points and circles
Jungle • Basis for coordinates
INPUT 2012 5

6. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
Administrative objects Human-made features
• No ligitations at boundaries • Made by humans
• Non-connex planar polygons (2D) – Parcels, buildings, streets, bridge, tunnels, etc.
• Often in hierarchical tesselations • Modeled by non-connex polygons (2D) or
– Countries, regions, provinces, cities polyhedra (3D)
– Natural parks • At some scale, roads are linear
• Total coverage of the globe
• At some scale, some objects can disappear
Object geometry Multiple representation
• Only one storing structure Street represented
by a graph
Street represented
by two polylines
• But many layout (mapping) structure issued Traffic
Cadaster
engineer
by generalization officer
• When layout geometry < threshold, then Street
Street
Technical
object will disapear maintenance
engineer
network
engineer
Street represented Street represented
by a surface by a volume
INPUT 2012 6

7. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
3 – Spatial relationships Spatial relations
• Conventional spatial relations • Topological (Allen, Egenhofer, Clementini,
etc.)
• New spatial relations
• Projective (cardinal)
• Distances
Egenhofer relations Projective relations
INPUT 2012 7

8. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
Main semantic properties
Examples of spatial relations of geographic concepts
Semantic properties
Scope Date
Agent
Duration
Coverage
Size
Shape Frequence
Location
Main semantic relations
Jungert operators (1/2)
of geographic concepts
Semantic Relations
SEPARATION
IS-A A
ADJACENCY
A<B center(A) < center(B) B
IS-PART-OF CONNECTIVITY
A=B center(A) = center(B) AB
HAS-PART OVERLAP
RELATIVE POSITION INTERSECTION
A|B Side by side A B
CONTAINMENT
SOURCE - DESTINATION
EXCLUSION Min(A) > Min (B)
A
PROXIMITY A%B Max(A) < Max (B)
SURROUNDNESS Length(A) < Length(B) B
DIRECTION EXTENSION
INPUT 2012 8

9. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
Jungert operators (2/2) Streets
Min(A) = Min (B) A
A[B Length(A) < Length(B) B
Max(A) = Max (B) A
A]B Length(A) < Length(B) B
Min(A) < Min (B) A
AB Length(A) ≤ Length(B) B
Sidewalk Lane Sidewalk
Max(A) > Max (B) A
A/B Length(A) ≤ Length(B) Lane
B
Motorway New spatial relations (2D)
• Horizontally
– Ribbon
Median – Relations
Lanes • End-to-end noted =
• Side-by-side noted II
Emergency lane
Verge/Shoulder
INPUT 2012 9

10. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
Motorway Sections
Shoulder
• Example: Motorway
Emer.
– Divided into « sections » End-to-end Lane
Section = = Ribbon Ribbon = =
– Each section divided into « ribbon » Side-by-side
Lane
= Median
Motorway = = Section
= Emerg.
Section = = II Shoulder II II Lane II Median
Lane
II
CONOSCENZE SPAZIALI E CONOSCENZE GEOGRAFICHE
New spatial relations (3D) Mountain
• Vertically
– layer
– Relations
• Bottom-to-top
• Top-to-bottom
• End-to-end
==> Geology
• Side-to-side
Buildings
R. Laurini
INPUT 2012 10

11. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
Vegetation layers 4 – GeoOWL
• Fuzzy layers • OWL: Ontology Web Language
• As 3D objects (bottom-to-top) • Geographic ontology based on OWL
• 8 million names for 6.5 million features
• As 2D surfaces (side-by-side)
• Modeling aspect
• Links with avec GeoNames
– Placenames (toponyms)
– Name of features (in English)
Key-concepts of GeoOWL GeoNames
<owl:Class rdf:about="&gml;/_Feature" /> <owl:ObjectProperty rdf:about="&gml;/exterior">
<owl:Class rdf:about="&gml;/_Geometry" />
<owl:Class rdf:about="&gml;/Point"> <owl:DatatypeProperty rdf:about="&gml;/lowerCorner">
<owl:Class rdf:about="&gml;/Polygon"> <owl:DatatypeProperty rdf:about="&gml;/upperCorner">
<owl:Class rdf:about="&gml;/Envelope"> <owl:DatatypeProperty rdf:about="&gml;/featurename">
<owl:Class rdf:about="&gml;/LineString"> <owl:DatatypeProperty rdf:about="&gml;/featuretypetag">
<owl:Class rdf:about="&gml;/LinearRing"> <owl:DatatypeProperty rdf:about="&gml;/relationshiptag">
<owl:ObjectProperty rdf:about="&geo;/where"> <owl:DatatypeProperty rdf:about="&geo;/point">
<owl:DatatypeProperty rdf:about="&geo;/line">
<owl:DatatypeProperty rdf:about="&gml;/pos"> <owl:DatatypeProperty rdf:about="&geo;/polygon">
<owl:DatatypeProperty rdf:about="&geo;/box">
<owl:DatatypeProperty rdf:about="&gml;/posList">
<owl:DatatypeProperty rdf:about="&geo;/elev">
<owl:Class rdf:about="&gml;/Point">
<owl:Class rdf:about="&gml;/LineString"> <owl:DatatypeProperty rdf:about="&geo2003;#lat">
<owl:Class rdf:about="&gml;/LinearRing"> <owl:DatatypeProperty rdf:about="&geo2003;#long">
INPUT 2012 11

12. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
Sub-classes Excerpt from the list
of geographic concepts
• ontology:A country, state, region
• DTCH ditch a small artificial watercourse dug for draining or irrigating the land
• ontology:H (water bodies) • DTCHD drainage ditch a ditch which serves to drain the land
• DTCHI irrigation ditch a ditch which serves to distribute irrigation water
• ontology:L (parks, areas) • DTCHM ditch mouth(s) an area where a drainage ditch enters a lagoon, lake or bay
• ESTY estuary a funnel-shaped stream mouth or embayment where fresh water mixes with sea water
• ontology:P city, village •
under tidal influences
FISH fishing area a fishing ground, bank or area where fishermen go to catch fish
• FJD fjord a long, narrow, steep-walled, deep-water arm of the sea at high latitudes, usually along
• ontology:R road, railroad mountainous coasts
• FJDS fjords long, narrow, steep-walled, deep-water arms of the sea at high latitudes, usually along
• ontology:S spot, buildings, farms, •
mountainous coasts
FLLS waterfall(s) a perpendicular or very steep descent of the water of a stream
• ontology:T mountain, hill, rock •
•
FLLSX section of waterfall(s)
FLTM mud flat(s) a relatively level area of mud either between high and low tide lines, or subject to
flooding
• ontology:U undersea • FLTT tidal flat(s) a large flat area of mud or sand attached to the shore and alternately covered and
uncovered by the tide
• ontology:V forest, heath • GLCR glacier(s) a mass of ice, usually at high latitudes or high elevations, with sufficient thickness to flow
away from the source area in lobes, tongues, or masses
• GULF gulf a large recess in the coastline, larger than a bay
GeoNames Ontology URI
• Ex. : for Cagliari, there are 2 URI
1: City:
http://www.geonames.org/maps/google_39.207_9.135.html
2: Province:
http://www.geonames.org/maps/google_39.245_9.091.html
– 1 : location
– 2 : information
INPUT 2012 12

13. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
http://sws.geonames.org/3020251/ Linked data
• Components of Italy (contains) :
http://www.geonames.org/3175395/contains.rdf
• Neighboring countries of Italy (neighbours) :
http://www.geonames.org/3175395/neighbours.rdf
• Geographic features (For example, the Eiffel Tower)
http://sws.geonames.org/6254976/nearby.rdf
http://www.geonames.org/3175395/neighbours.rdf GeoNames Services
• 1 astergdem • 18 gtopo30
• 2 children • 19 hierarchy
• 3 cities • 20 neighbourhood
• 4 countryCode • 21 neighbours
• 5 countryInfo • 22 ocean
• 6 countrySubdivision • 23 postalCodeCountryInfo
• 7 earthquakes • 24 postalCodeLookup
• 8 extendedFindNearby • 25 postalCodeSearch
• 9 findNearby • 26 rssToGeo
• 10 findNearbyPlaceName • 27 search
• 11 findNearbyPostalCodes • 28 siblings
• 12 findNearbyStreets • 29 srtm3
• 13 findNearByWeather • 30 timezone
• 14 findNearbyWikipedia • 31 weather
• 15 findNearestAddress • 32 weatherIcao
• 16 findNearestIntersection • 33 wikipediaBoundingBox
• 17 get • 34 wikipediaSearch
INPUT 2012 13

14. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
5 – Gazetteers Examples
• Byzantium, Constantinople, Istanbul
• Gazetteer = dictionary of placenames
• Roma of Romulus and of today
• What to store:
• Cagliari, city or province
– Names with variants along time
• Etc
– Names in different languages
– Geometry (varying)
– Feature type Many-to-many
– Neighbors Placenames Places
– Rivers: confluence
Geonaming Generic tables
• From the coordinates of a place, assigning a name to Names (idn, text)
this place Geographic
– line Category (idc, category-type) Ontology
– area
Geometry (idn, idc, date, not-connected-polygon)
• Problems of linguistics
– multilingual problem
Other-names (idn1, idn2, type, language)
• What name?
– Name in the official language of the country
– Name in the language of the user
– Name in the language of the system
INPUT 2012 14

15. Geographic Ontologies and Spatial Relationships Prof. Robert Laurini
For municipalities For countries
Belongs-to (idn, date, country-name) Population (idn, date, poulation)
Population (idn, date, poulation) Neighbours (idn, (neighbouring-countries)*)
Neighbours (idn, (neighbouring-municipalities)*)
For rivers
Confluence (idn, main-confluent-river-idn)
6 – Conclusions
• Ontologies as tools
• For interoperability
• For clarifying vocabulary
• Difficulties to properly define geographic features
• Semantically
• Geometrically
• Topologically
• Importance of spatial relationships
• Possibly of using new spatial relations
• Links with gazetteers
• Geographic ontologies (with spatial relationships)
INPUT 2012 15