146
Omar Boucelma, Mehdi Essid, and Yassine Lassoued
Length > 1000. A join is then performed between the results of the prefix and the
suffix queries in order to build the final result. Figure 7.11 illustrates the execution
plan of query query0 of the GEP.
7.6 VirGIS/Q: A Quality Geographical Mediator
In this section, we describe how to handle quality parameters in answering a mediated query that may involve many data sources with different qualities.
7.6.1 Problem Sketch
Consider a user who is looking for highways of a given region b that satisfy a given
quality condition. The user may express a SQL like query Q 0 :
Q 0 : Select all information about road objects of type Highway such that data
cover region b and have been validated during the last 10 years (e.g. 1995–2005),
with a deficit rate of road objects that is less than 1%, a geometric accuracy of at
most 5 meters and a high semantic accuracy of the operational state of the roads.
First of all, query processing may be different depending on the choice of the spatial
cover b. For example, if we respectively choose region B, B
, or B
of Fig. 7.12, the
number of data sources that will participate in query processing is respectively two,
three, or zero. The case of B
is trivial since the query has no answer. Examples with
B and B
are more complicated for the reasons explained below.
Case of Region B: Both TQh and NT are eligible over the whole surface B. Both
sources satisfy the condition for the validation date. The deficit rate of the first source
(1%) is sufficient; that of the second is higher (i.e. worse) than required (5% > 1%);
however, we cannot reject the data source because when considered with the first
2
B’
B’ 3
B’’
B’ 1
B’ 4
BDTQ−31hh202
BDTQ−21hh201
B
B’
NTDB−21e05
Sherbrooke
Fig. 7.12. Two sources participate with B as spatial cover, three sources with B
, and none
with B
Omar Boucelma, Mehdi Essid, and Yassine Lassoued
Length > 1000. A join is then performed between the results of the prefix and the
suffix queries in order to build the final result. Figure 7.11 illustrates the execution
plan of query query0 of the GEP.
7.6 VirGIS/Q: A Quality Geographical Mediator
In this section, we describe how to handle quality parameters in answering a mediated query that may involve many data sources with different qualities.
7.6.1 Problem Sketch
Consider a user who is looking for highways of a given region b that satisfy a given
quality condition. The user may express a SQL like query Q 0 :
Q 0 : Select all information about road objects of type Highway such that data
cover region b and have been validated during the last 10 years (e.g. 1995–2005),
with a deficit rate of road objects that is less than 1%, a geometric accuracy of at
most 5 meters and a high semantic accuracy of the operational state of the roads.
First of all, query processing may be different depending on the choice of the spatial
cover b. For example, if we respectively choose region B, B
, or B
of Fig. 7.12, the
number of data sources that will participate in query processing is respectively two,
three, or zero. The case of B
is trivial since the query has no answer. Examples with
B and B
are more complicated for the reasons explained below.
Case of Region B: Both TQh and NT are eligible over the whole surface B. Both
sources satisfy the condition for the validation date. The deficit rate of the first source
(1%) is sufficient; that of the second is higher (i.e. worse) than required (5% > 1%);
however, we cannot reject the data source because when considered with the first
2
B’
B’ 3
B’’
B’ 1
B’ 4
BDTQ−31hh202
BDTQ−21hh201
B
B’
NTDB−21e05
Sherbrooke
Fig. 7.12. Two sources participate with B as spatial cover, three sources with B
, and none
with B