7 A Quality-enabled Spatial Integration System
147
one it will improve (decrease) the total deficit rate. According to the quality conditions, the geometric property must be extracted from TQe, information about the
operational state must be extracted from NT, while rest of the information can be
extracted from any source.
Note that in modifying the quality conditions we may also modify the origins of
attributes. For example, requiring a validation date between 1996 and 2004 instead
of requiring a geometric accuracy of at most 5 m implies that the geometric property
should be extracted from NT instead of TQe.
Suppose now that the deficit rate required by the user is 0.5% (instead of 1%). In
this case, even though both data sources do not satisfy the condition with the deficit
rate, they are still interesting and, if used together, they may lead to a good result. In
fact, the probability that an object is absent from both data sources is 1% × 5%, that
is 0.05%, which is less than 0.5%.
Case of Region B
: In region B
, TQh and NT cover different zones. Zones B
1 , B
2 ,
B
3 , and B
4 do not involve the same data sources. However, each of them is homogeneous and can be treated in the same way we did above for B and B
.
7.6.2 Requirement Analysis
The goal of our (quality) mediation system is to allow a community of users to share
a set of heterogeneous, autonomous geographical data sources of various qualities.
A user poses a query (against the mediated schema) and asks for geographical objects of a particular kind, with a specific quality (conditions over the spatial cover,
the geometric accuracy, the date of validity, the deficit and surplus rates, etc.). The
system must explore the different data sources with their information quality in order
to retrieve interesting data from appropriate sources and integrate them to generate,
if possible, a result with an acceptable quality.
In order to achieve the quality mediation goal, we need to take into account many
parameters:
1. Geometry: because geographical objects have geometric properties (punctual,
linear, polygonal, etc.).
2. Spatiality: since data are linked to the ground and cover different zones, depending on the query zone the data sources used by the system may be different (see
Sect. 7.6.1, for example).
3. Heterogeneity in data representation, different classifications, and different
schemas.
4. Data Quality: its definition (model) and the requirements it imposes on the
schema mapping language, the query model, and the query rewriting
mechanism.
7.6.3 Solution Sketch
The mediation system uses a set of mapping rules that describe correspondences
between global and local sources. A mapping rule associates pairs of corresponding elements (from the global and a source schema) and quality information. These
147
one it will improve (decrease) the total deficit rate. According to the quality conditions, the geometric property must be extracted from TQe, information about the
operational state must be extracted from NT, while rest of the information can be
extracted from any source.
Note that in modifying the quality conditions we may also modify the origins of
attributes. For example, requiring a validation date between 1996 and 2004 instead
of requiring a geometric accuracy of at most 5 m implies that the geometric property
should be extracted from NT instead of TQe.
Suppose now that the deficit rate required by the user is 0.5% (instead of 1%). In
this case, even though both data sources do not satisfy the condition with the deficit
rate, they are still interesting and, if used together, they may lead to a good result. In
fact, the probability that an object is absent from both data sources is 1% × 5%, that
is 0.05%, which is less than 0.5%.
Case of Region B
: In region B
, TQh and NT cover different zones. Zones B
1 , B
2 ,
B
3 , and B
4 do not involve the same data sources. However, each of them is homogeneous and can be treated in the same way we did above for B and B
.
7.6.2 Requirement Analysis
The goal of our (quality) mediation system is to allow a community of users to share
a set of heterogeneous, autonomous geographical data sources of various qualities.
A user poses a query (against the mediated schema) and asks for geographical objects of a particular kind, with a specific quality (conditions over the spatial cover,
the geometric accuracy, the date of validity, the deficit and surplus rates, etc.). The
system must explore the different data sources with their information quality in order
to retrieve interesting data from appropriate sources and integrate them to generate,
if possible, a result with an acceptable quality.
In order to achieve the quality mediation goal, we need to take into account many
parameters:
1. Geometry: because geographical objects have geometric properties (punctual,
linear, polygonal, etc.).
2. Spatiality: since data are linked to the ground and cover different zones, depending on the query zone the data sources used by the system may be different (see
Sect. 7.6.1, for example).
3. Heterogeneity in data representation, different classifications, and different
schemas.
4. Data Quality: its definition (model) and the requirements it imposes on the
schema mapping language, the query model, and the query rewriting
mechanism.
7.6.3 Solution Sketch
The mediation system uses a set of mapping rules that describe correspondences
between global and local sources. A mapping rule associates pairs of corresponding elements (from the global and a source schema) and quality information. These