160
Paola Podest` a, Barbara Catania, and Alberto Belussi
resolutions, leading to accuracy, precision, and consistency issues. On the other side,
from a user point of view, multiresolution may result in a gap between the available
data and the user’s knowledge of such data during query specification, reducing user
satisfaction in using a given application.
A possible approach to deal with such heterogeneity would be that of extending
spatial data integration solutions to cope with specific data quality parameters. As
we have seen in Chap. 7, this may lead to the definition of specific quality-aware
query languages. However, when the user does not know exactly the schema of the
data set to be queried, as often happens in distributed environments, this solution
may not work. Another approach consists in tackling the problem from the top, by
extending query specification and processing to directly exploit multiresolution. For
this purpose, in Chap. 4, an overview of methods for the progressive transmission
of multiresolution spatial data over the Web has been presented. In this chapter, we
consider a complementary problem concerning how query processing techniques, in
a distributed environment, have to be revised in order to use information concerning
multiresolution maps, thus improving the quality of query processing results. The
proposed techniques focus on three main GIS application contexts:
1. Similarity-based Processing. Due to the presence of multiresolution data sets
over distributed architectures, the user may not know exactly the spatial domain
she wants to query in terms of properties, available features, and geometric feature types. This gap may impact the quality of the results obtained by a query
execution, reducing user satisfaction in using a given application as the result
obtained may not exactly correspond to user needs.
2. GIS Mediation Architectures. Mediation systems provide users with a uniform
access to a multitude of data sources via a common model, without duplicating such data. The user poses their query against a virtual global schema and
the query is in turn rewritten into queries against the real local sources, taking
into account differences in the data models and query languages. Results obtained from the various sources are then merged and returned to the user (see
Chap. 6 for a review of geographical information fusion techniques). In the context of GIS data, VirGIS is a mediation system based on open geospatial consortium (OGC) standards that addresses the issue of integrating GIS data and
tools [3, 4, 20] (see Chap. 7). In general, mediator systems, including VirGIS,
take into account differences concerning feature representation in local sources.
Differences in data sources may depend on how each single data source models
spatial objects in terms of their descriptive attributes (length of a river, population in a town), their geometric type (region, line, point), and their spatial
relations. Unfortunately, mediators usually do not consider the impact of spatial
relationship information on query rewriting. The problem here is that, since not
all spatial relationships are defined for any pair of geometric types, depending
on object types in each local source schema, different spatial predicates should
be considered for execution at the local level in order to return results consistent
with the global request.
Paola Podest` a, Barbara Catania, and Alberto Belussi
resolutions, leading to accuracy, precision, and consistency issues. On the other side,
from a user point of view, multiresolution may result in a gap between the available
data and the user’s knowledge of such data during query specification, reducing user
satisfaction in using a given application.
A possible approach to deal with such heterogeneity would be that of extending
spatial data integration solutions to cope with specific data quality parameters. As
we have seen in Chap. 7, this may lead to the definition of specific quality-aware
query languages. However, when the user does not know exactly the schema of the
data set to be queried, as often happens in distributed environments, this solution
may not work. Another approach consists in tackling the problem from the top, by
extending query specification and processing to directly exploit multiresolution. For
this purpose, in Chap. 4, an overview of methods for the progressive transmission
of multiresolution spatial data over the Web has been presented. In this chapter, we
consider a complementary problem concerning how query processing techniques, in
a distributed environment, have to be revised in order to use information concerning
multiresolution maps, thus improving the quality of query processing results. The
proposed techniques focus on three main GIS application contexts:
1. Similarity-based Processing. Due to the presence of multiresolution data sets
over distributed architectures, the user may not know exactly the spatial domain
she wants to query in terms of properties, available features, and geometric feature types. This gap may impact the quality of the results obtained by a query
execution, reducing user satisfaction in using a given application as the result
obtained may not exactly correspond to user needs.
2. GIS Mediation Architectures. Mediation systems provide users with a uniform
access to a multitude of data sources via a common model, without duplicating such data. The user poses their query against a virtual global schema and
the query is in turn rewritten into queries against the real local sources, taking
into account differences in the data models and query languages. Results obtained from the various sources are then merged and returned to the user (see
Chap. 6 for a review of geographical information fusion techniques). In the context of GIS data, VirGIS is a mediation system based on open geospatial consortium (OGC) standards that addresses the issue of integrating GIS data and
tools [3, 4, 20] (see Chap. 7). In general, mediator systems, including VirGIS,
take into account differences concerning feature representation in local sources.
Differences in data sources may depend on how each single data source models
spatial objects in terms of their descriptive attributes (length of a river, population in a town), their geometric type (region, line, point), and their spatial
relations. Unfortunately, mediators usually do not consider the impact of spatial
relationship information on query rewriting. The problem here is that, since not
all spatial relationships are defined for any pair of geometric types, depending
on object types in each local source schema, different spatial predicates should
be considered for execution at the local level in order to return results consistent
with the global request.