Querying a Geographic Database using an OntologyBased Methodology

1
Querying a Geographic Database using an
Ontology-Based Methodology

• Renata Viegas
• Valéria G. Soares

valeria_at_di.ufpb.br renata_at_ppgsc.ufrn.br
2
Summary

• Motivation
• System Architecture
• Semantic Layer
• Application Domain
• Conclusion

3
Motivation

• GIS are multidisciplinary systems
• Geographic features are collected and stored in
  GIS that were modeled based on a specific
  researcher vision.
• Current GIS must be able to solve the semantic
  interoperability.

4
Motivation

• Semantic interoperability
• A geographic feature could have more than one
  description
• Interoperability based on the use of ontologies
  as being a knowledge database type.

5
Motivation

• Geographic Ontologies
• A geographic ontology is a conceptualization of a
  phenomenon or geographic object in the real
  world.
• Characteristics of geographic objects must be
  embodied to the ontology
• Location
• Topology
• Direction

6
Geographic Ontologies

• Relationship through ontologies classes with
  typical geographic relationships

São José dos Campos
Campos do Jordão
Near
Within
João Pessoa
Northeast Region
7
Summary

• Motivation
• System Architecture
• Semantic Layer
• Application Domain
• Conclusion

8
System Architecture

• Our Problem
• Different professionals of distinct research
  areas want to access the same GDB
• Our solution
• Based on geographic ontologies

9
System Architecture

• The semantic layer intermediates the users
  queries with the geographic database
• Each one of the users community could interact
  with the system using only specific terms of its
  research area, and could receive its queries
  answers in an appropriate way.

10
General System Architecture
11
Query Process

• The user query will be submitted
• The system will recognize the used terms.
• The next step is to look for the terms and
  concepts used in this query, in the users
  ontology
• comparing the ontology with the GDB ontology,
  looking for equivalent concepts.

12
Query Process
QUERY USER
Query Management
Database Request
PHP
Users, Concepts, Terms
Semantic Layer
Java / Jena
Users Ontology
GDB Ontology
13
Detailing the Semantic Layer

• Step 1 Define the ontologies
• Step 2 Bind definition between the classes of
  the different ontologies
• The result of this binding process is a formal
  structure with expressions that show which terms
  of each ontology is related to others terms of
  another ontology.

14
Detailing the Semantic Layer

• Mapping synonyms classes
• Synonym classes are classes whose concepts have
  the same meaning, independent of their given
  names, that are related of the specific knowledge
  of each community.
• Ontologies Manipulation
• We use Jena API to generate graphs RDF, which is
  represented by resources, properties and
  literals
• From the Jena API methods we can manipulate and
  compare the ontologies.

15
Semantic Layer
16
Semantic Layer Modules

• Users' Management Module
• The first step to submit a query is to inform
  which type of user wants to interact with the
  system
• The system will show a pay-define queries
  interface, with only specific terms of this type
  of users, based on the defined ontologies.

17
Semantic Layer Modules

• Ontologies Management Module
• Activate ontologies
• User ontology and the ontology that represents
  the contents of the GDB
• The ontologies are stored in ontologies server,
  and are accessed through their URLs
• Methods of the Jena API will be used to construct
  the graphs (models) of the ontologies.

18
Semantic Layer Modules

• Query Preparation Module
• This module will identify and store the key terms
  of the query

19
Semantic Layer Modules

• Comparative Module
• Search of similar terms in the ontologies
• The similarity is defined manually, based on the
  interviews with professionals of the different
  areas
• The OWL Tags
• equivalentClass and sameAs

20
Semantic Layer Modules

• Comparative Module
• The ltowlsameAsgt tag is used when we have
  different nomenclatures that refers to a same
  entity
• A typical use of owlsameAs is for ontologies
  unification, to say that two individuals classes,
  defined in different documents, are equals.

21
Semantic Layer Modules

• Comparative Module
• Methods of the Jena API treat the similarities
  binding classes of an ontology to another one
• The getSameAs() method, by the OntResource
  interface, is used to find the similar classes in
  the ontologies.

22
Semantic Layer Modules

• Query Generation Module
• Mount the query that will be submitted to the
  database
• Use the terms found in the search for the
  similarity, as well as the relationship used in
  the query interface.

23
Semantic Layer Modules

• Dictionary Generation Module
• Mount a detailed text, with the key terms of the
  query, supplying to the user descriptions about
  the geographic features involved in his queries.

24
Summary

• Motivation
• System Architecture
• Semantic Layer
• Application Domain
• Conclusion

25
Application Coral Reef Domain

• Geographic Domain Coral Reefs
• We have three different communities geologists,
  biologists and tourists
• Development of three different ontologies for
  each one of these communities.
• The ontologies give support to the construction
  of adaptable interfaces for each community.

26
Developed Ontologies

• Geologist Ontology

27
Developed Ontologies

• We also have developed ontology for the
  biologists community and tourists community
• The geologist ontology is the same of the
  geographic database ontology (GDB).

28
Prototype Query Example

• Example A tourist wants to find the best area
  for dip in the sea nearby the coral reefs
• Activated ontologies for this query the
  geologist (Ogeo) and the tourist (Otur)

This information is not stored on our database
29
Prototype Query Example - Tourist

• Steps
• Look for the term dip in the sea on the tourist
  ontology
• We defined in the Otur that a tourist could dip
  in the sea around the floating boats or within
  the natural pools

30
Prototype Query Example - Tourist

• Look for an equivalent term on the geologist
  ontology
• No similar class to dip in the sea could be
  found in the Ogeo at this point

31
Prototype Query Example - Tourist

• Steps (Continue)
• Go down one more step on the tourist ontology
• Verify if there is more information in the
  relationships between classes or subclasses
• dip in the sea appears in the tourist ontology
  with properties that bind this class to other
  ones in the ontology
• The property within binds the class dip in the
  sea (domain) to the class natural pool (range)
• The property "surround", binds the class dip in
  the sea (domain) to the class floating boats
  (range)

32
Prototype Query Example - Tourist

• Steps (Continue)
• the system will search now for classes similar to
  natural pool and floating boats on the
  geologist ontology
• The class floating boats is defined on both the
  tourist ontology Otur and the geologist ontology
  Ogeo
• The class natural pool is not found

33
Prototype Query Example - Tourist

• Steps (Continue)
• The Comparative Module goes down one more step in
  the graph generated from the tourist ontology,
  looking for relationships between classes
• The relationship is found Natural pools
  surround Coral Reef

34
Prototype Query Example - Tourist

• Steps (Continue)
• The query that will be submitted to the GDB is
  show all the areas that surround the floating
  boats and the coral reefs bodies

SELECT buffer (flutuante.flutuante_geom, 10) AS
flutuante_geom, buffer(corpo_coralineo.geom_cabe
co, 10) AS geom_cabeco FROM flutuante,
corpo_coralineo AS foo USING UNIQUE oid USING
SRID -1
35
Results
36
Results
37
Prototype Query Example - Biologist

• A biologist wants to know where he could find no
  consolidate substrates on the coral reef region
• Activated ontologies for this query the
  biologist (Obio) and the geologist (Ogeo)

38
Prototype Query Example - Biologist

• Steps
• Look for the no-consolidate substrate class in
  the Obio ontology
• Look for class similarities on the definition of
  this class
• The class no-consolidate substrate in the
  biologist ontology has no similar class in any
  other ontology
• The comparative module will go down one more
  level, verifying whether the no-consolidate
  substrate class has some relationships with
  others classes or has sub-classes

39
Prototype Query Example - Biologist

• Steps
• The comparative module finds that the
  no-consolidate substrate class has sub-classes
  (relationship ISA) with the classes sand,
  gravel and mud

40
Prototype Query Example - Biologist

• Look for classes similar to sand, gravel and
  mud on the geologist ontology

41
Prototype Query Example - Biologist

• Steps
• The conceptual similarity is found
• The Query Generator Module will construct this
  following SQL query clause

SELECT geom_areia AS geom_areia, geom_lama AS
geom_lama, geom_cascalho AS
geom_cascalho FROM areia, cascalho, lama AS foo
USING UNIQUE oid USING SRID -1
42
Results
43
Results
44
Summary

• Motivation
• System Architecture
• Semantic Layer
• Application Domain
• Conclusion

45
Conclusions

• Summary of Contributions
• Development of the ontologies for the coral reef
  domain, based on three points of view of
  different communities the geologists community,
  the biologists and the tourists
• We choose a natural environment of coral reefs
  because no other proposal makes use of semantic
  terms with this kind of geographic databases.

46
Conclusions

• Summary of Contributions
• An ontology-based mechanism allows different
  users communities, through geographic
  ontologies, to access the same database, without
  knowing its internal structure
• With only one database implementation and the
  definition of different communities ontologies,
  anyone can search the database, in a transparent
  way, using a specific interface.

47
Conclusions

• Summary of Contributions
• A Framework for this mechanism
• The Semantic Layer can be adapted to any spatial
  domain of multidisciplinary interest.
• Future Work
• Extend this Architecture
• Bind the classes automatically
• Define ranking of similarities between the
  classes.

48
Querying a Geographic Database using an
Ontology-Based Methodology

• Renata Viegas
• Valéria G. Soares

valeria_at_di.ufpb.br renata_at_ppgsc.ufrn.br