6
Alberto Belussi, Barbara Catania, Eliseo Clementini, and Elena Ferrari
geographical and, more generally, of the spatial context. Therefore, specific solutions
have to be provided for them.
Data integration aims at overcoming problems concerning data conflicts to provide homogeneous access to local sources. Data integration solutions rely on the
identification of different objects, represented in distinct sources and related by some
semantic link, and on the resolution of conflicts existing between such objects.
In order to describe the semantics of each data set, a schema or, more generally,
an ontology can be used. An ontology can be defined as “an explicit specification of a
conceptualization” [7]. An ontology, besides describing the structural characteristics
of a data set, i.e. its schema, also provides logical systems to be used for defining and
reasoning about relationships and constraints existing between the data set concepts.
The process of identifying the relationships between corresponding elements in two
heterogeneous ontologies is often called ontology alignment [18]; on the other hand,
the process of constructing a single combined ontology based on the identified relationships, and therefore based on a given ontology alignment, is called “ontology
integration” and, more specifically, geographical information fusion [5]. Ontology
alignment solutions can be intensional, if they are based on concepts definitions (e.g.
properties of roads), or extensional, if they consider concepts instances (e.g. the single roads in the data set).
Among the existing approaches for ontology integration, mediation has been
investigated in depth in the past few years [24]. A mediation system can be defined as a system providing to the user a uniform interface by which different
data sources can be accessed through a common global model. The mediation
system relies on rules for mapping concepts in the global schema into concepts
contained in each local source based on a given and predefined ontology alignment. The architecture of a mediation system is based on two main components: the
mediator and the wrappers. The mediator allows semantic translations by rewriting,
using the predefined mapping rules, the user’s query into queries over data sources
expressed in a common query language, which is specific to the mediator. Each
data source is accessed through a wrapper. When a query is posed against a data
source, the corresponding wrapper translates it according to the data source query
language.
In order to increase interoperability, OGC has defined several standards for geographical data representation, transfer, and access to be used in mediator systems
and in distributed architectures in general. OGC has adopted GML for the XML representation and transport of geographical data [16] and Web Feature Services (WFS)
for describing or getting features from a spatial data source on the Web [17]. Although such standards can help the development of interoperable applications, they
do not provide solutions to ontology alignment and integration problems, which must
be addressed in order to provide adequate data integration approaches.
Another relevant integration issue concerns the quality of the accessed data.
Indeed, high quality at each local source may correspond to low quality at the integration level. For geographical data, several models have been proposed by various
organizations, converging to an ISO/TC 211 model [11, 12]. Each model proposes
Alberto Belussi, Barbara Catania, Eliseo Clementini, and Elena Ferrari
geographical and, more generally, of the spatial context. Therefore, specific solutions
have to be provided for them.
Data integration aims at overcoming problems concerning data conflicts to provide homogeneous access to local sources. Data integration solutions rely on the
identification of different objects, represented in distinct sources and related by some
semantic link, and on the resolution of conflicts existing between such objects.
In order to describe the semantics of each data set, a schema or, more generally,
an ontology can be used. An ontology can be defined as “an explicit specification of a
conceptualization” [7]. An ontology, besides describing the structural characteristics
of a data set, i.e. its schema, also provides logical systems to be used for defining and
reasoning about relationships and constraints existing between the data set concepts.
The process of identifying the relationships between corresponding elements in two
heterogeneous ontologies is often called ontology alignment [18]; on the other hand,
the process of constructing a single combined ontology based on the identified relationships, and therefore based on a given ontology alignment, is called “ontology
integration” and, more specifically, geographical information fusion [5]. Ontology
alignment solutions can be intensional, if they are based on concepts definitions (e.g.
properties of roads), or extensional, if they consider concepts instances (e.g. the single roads in the data set).
Among the existing approaches for ontology integration, mediation has been
investigated in depth in the past few years [24]. A mediation system can be defined as a system providing to the user a uniform interface by which different
data sources can be accessed through a common global model. The mediation
system relies on rules for mapping concepts in the global schema into concepts
contained in each local source based on a given and predefined ontology alignment. The architecture of a mediation system is based on two main components: the
mediator and the wrappers. The mediator allows semantic translations by rewriting,
using the predefined mapping rules, the user’s query into queries over data sources
expressed in a common query language, which is specific to the mediator. Each
data source is accessed through a wrapper. When a query is posed against a data
source, the corresponding wrapper translates it according to the data source query
language.
In order to increase interoperability, OGC has defined several standards for geographical data representation, transfer, and access to be used in mediator systems
and in distributed architectures in general. OGC has adopted GML for the XML representation and transport of geographical data [16] and Web Feature Services (WFS)
for describing or getting features from a spatial data source on the Web [17]. Although such standards can help the development of interoperable applications, they
do not provide solutions to ontology alignment and integration problems, which must
be addressed in order to provide adequate data integration approaches.
Another relevant integration issue concerns the quality of the accessed data.
Indeed, high quality at each local source may correspond to low quality at the integration level. For geographical data, several models have been proposed by various
organizations, converging to an ISO/TC 211 model [11, 12]. Each model proposes