a geo-ontology to support the semantic integration of geoinformation from the national spatial data...
TRANSCRIPT
A geo-ontology to support the semantic integration of
geoinformation from the National Spatial Data Infrastructure
Authors:
Paulo J. A. Gimenez, Mastering Asterio Kiyoshi Tanaka, DSc
Fernanda Baião, DSc
Presentation
ContextualizationIncreasingly demand for the availability of integrated
geoinformation: The profusion of geoinformation. The diversity of providers. Over 30 years of evolution and growth of the geospatial content.
The National Spatial Data Infrastructure (INDE) : To provide an effective and efficient access and availability of spatial data. Governments and public entities engaged.
Web 2.0 technologies: Mature and available.
The necessity for formalization, representation and integration of the geoinformation at the semantic level: still open !!! [Diaz ET AL 2012]
2
The Challenge of Geo-SemanticInteroperability (by SDI dimension) and Integration (by Data
dimension).Main Levels of Interoperability
Syntactic: how geographic objects will move among institutions and these to society? Approaches from ISO, OGC and OASIS patterns and directives.
Semantic: the choice of an ontology or use of semantic resources to describe geographic objects and their interrelationships.
Main Geospatial heterogeneity issues Geospatiality: features and geometries of features; geographic and non-
geographic relationships; systems and coordinate; scales conflicts; Geo-semantics: discernment of a feature; spatial reasoning and
representation dissonance.
3
Goal of this proposalPresent
an approach to the creation of a geo-ontology for representing geographic objects
within the context of NSDI (INDE)from the associated standards
enabling the discovery and integration of geoinformation.
4
ET-EDGV Feature concepts
ET-ADGV Feature constructions
MGB Profile Feature/ Collection metadata
BNGB Geographical names
The Brazilian Context: “INDE”
5
The NSDI (INDE) Arrange of SDI in
different levels and providers.
Cartographic and basic geoinformation structured by ET-EDGV standard.
Follows e-PING architecture, references the OGC standards and adds owned standards: MGB Profile (Brazilian
Geographic Metadata Profile), based on ISO 19115:2003.
ET-EDGV (Technical Specification for Geographic Vector Data Structure).
[CONCAR 2010]
Geo-ontology basic conceptsGeo-ontology = {C, R, A, X, I}, where:
C (concepts) represents concept set of geographic object;R (relations) is a relation set and it mainly describes relation set among
concepts;A (attribute) shows attribute’s set of geographic objects;X (axioms) is a lot of axioms and it is a constraint rules among the concepts,
relation and attribute; I (instances) is a set of definition about instances.
[Wang, Li and Song 2008]
7
Methodological approach usedCombining existing methodologies from the literature:
The Simple Knowledge-Engineering Methodology [Noy and McGuinness 2001]. Simplicity and efficiency. Uses a set of competency questions (CQ) to determine the scope of the ontology.
The approach of Wang ET AL [Wang, Li and Song 2008]
Uses the "Concept lattice“: Sets of objects and attributes from geographic concepts that represent the main aspects
of the geospatial domain: is-a, kind-of, part-whole, dependency, instantiation and member relationships, as well as the relationship between attributes and concepts.
The work from Bishr [Bishr 2008]
Specifically addresses the geographical domain. Establishes the elements that must be observed during the construction of the
geo-ontology concepts.
8
Methodological approach used
Following the steps and directives:
9
Simple Knowledge-Engineering Complementing with Wang’s approach
Considering main elements in Bishr’s work
Uses Competency Questions to define the scope
Confirm the scope of geo-ontology The context: establish the set of assertions and conditions considering a restricted vocabulary and spatial-temporal perspectives
Determine the domain and scope
Reuse existing ontologies
List important terms in the ontology
List ontological properties for the geographical concept
Identity criteria: establish sufficient conditions to determine the identity of a concept, organize the taxonomy of concepts and persist in time.
Spatial reference system . Mereotopology. Limits: “bona-fide“ and "fiat" . The shape and size.
Define classes and class hierarchies
Ensure the relationship between geographical concepts
Define class properties Collect concepts meaning, their attributes, images and instances
Define facets of these properties
Create instances Build the prototype model / geospatial ontology system
The geo-ontology of INDE
10
Reuse and reference to known models of the geographical
Ontology for Geographical Names
11
The Geographic name or toponym standard allows the identification of a Geographic Feature or Accident.
The toponymic phrase consists of two parts: Generic term: the element on the geographical entity that receives the name. Specific term: the element that distinguishes the identity of the geographic
element.[IBGE 2010, Lima 2011 apud Dick 1990]
The proposed ontology brings: Correlation between term and toponyms. Lists of generic terms (as proposed in [IBGE 2010]). Support for names denoting variation in gender (male x female) and number (plural x
singular). Alternative toponymic phrases for a particular geographical area. Support for geographical names composed of multiple toponymic phrases.
Ontology for Geographical Names
12
Ontology for Brazilian Geographic Metadata Profile
13
Ontologies for geographic metadata add semantic meaning and relationship to describe the underlying data.[Di and Zhao 2008]
The proposed ontology describes: The basic characteristics for the representation of concepts defined in the
MGB Profile and related to the “ISO 19115” ontology. Sections and entities: as ontology classes. Information and elements: as properties of classes or enumerated lists. Same information elements shared among several sections with distinct
terminology: as synonym relationships. Cardinality restrictions mapped according to:
Ontology for Brazilian Geographic Metadata Profile
14
Ontology for Brazilian Geographic Domain
15
Based on the characteristics of geographic objects, spatial relationships and spatial primitives described in ET-EDGV. The guidelines for the construction of each element and concept defined
in ET-ADGV.The proposal ontology describes:
Classes and Relationship evaluated at the semantic level: as concepts based on class definitions and subtypes described.
Hierarchical classification of concepts: as the categories of ET-EDGV and classifications of geo-concepts proposed by Wang [Wang, Li and Song 2008].
The conceptual connection with the MGB Profile ontology. Cardinality restrictions and Conditionality relations mapped according to:
Ontology for Brazilian Geographic Domain
16
*Partial sub-ontology
The Competency Questions validation
17
CQ1: Which conditions or characteristics are required by a Geographic name so that it addresses (identifies) a Geographic Feature?
CQ2: Which conditions or characteristics are required by a Geographic Metadata so that it can be associated to a Geographic Name when identifying Geographic Feature?
CQ3: How are the needs for cartographic generalization of geographic features be identified?
CQ4: How can we identify the same object being represented as distinct cartographic features using different scales?
Application of geo-ontology proposed
18
The proposed geo-ontology must be considered to: Integration of the associated metadata. Correlation with the geographical names. The alignment with the ontology that describes the INDE geo-
services.The example scenario considers:
Integrated geoinformation is obtained from the basic geoinformation available in INDE.
Implementation of geo-processing to compose the integrated geo-information with its resulting metadata.
Related Works
19
Some studies have been made to define and specify the possible structuring of geo-ontologies sets to represent geographic space: [Bishr 2008],[Kun, Wang and Shuang-Yun 2005] and especially
Di and Zhao [Di and Zhao 2008] defines several levels of
abstraction for geo-ontologies
Wang's approach [Wang, Li and Song 2008] uses the "Concept lattice"
Final Considerations
20
The main contribution of this proposal: The proposed geo-ontology
The combination of concepts from the geographic names, metadata and geographic entities, providing support for analysis, applications and multifaceted uses.
Secondary effects: Intention to (re)open the discussion and the application prospect to maximize
the use of basic geoinformation available in INDE.
This work is essential part of study about semantic integration of several heterogeneous sources in the Brazilian context. Much has to be done yet, to achieve a geo-ontology that can be accepted as
the basis for this main study.
Final Considerations
21
Some future work in order to mature the geo-ontology proposed: Extension to
cover the needs of Systematic Cadastral Mapping as soon as the ET-EDGV specifies them;
represent metadata of geo-services that are not yet covered by the MGB Profile and adaptation of coded values lists to reflect the national context;
geo-field in alignment to ET-PCDG under elaboration; Creation of geographic quality control ontology for validation and verification
of geospatial data quality for alignment with the future ET-CQPCDG specification;
Expansion of Brazilian Geographic Domain Ontology to match ET-EDGV specification in a
complete way, considering all rules and orientations in there; Brazilian Geographic Names Ontology to treat the concepts associated with
historical, ethnological and linguistic characteristics of toponyms.
Implementation of a framework for semantic integration of spatial data from INDE using this proposed geo-ontology: under development.
A geo-ontology to support the semantic integration of geoinformation from the
National Spatial Data Infrastructure
[email protected]@uniriotec.br
Congratulations