Database Reference
In-Depth Information
R1
R4
R5
R2
R6
R3
Figure 4.6 The double-counting problem.
a cell
C
x,y,t
, the measures
crossX
and
crossY
give the number of distinct
trajectories crossing the spatial borders between
C
x,y,t
and
C
x
+
1
,y,t
and
C
x,y
+
1
,t
,
respectively. Analogously,
crossT
gives the number of distinct trajectories
crossing the temporal border between
C
x,y,t
and
C
x,y,t
+
1
. Knowing the values
of presence for two adjacent cells,
C
x,y,t
and
C
x
+
1
,y,t
, the aggregate value of
pres
over a new cell
C
x
,y
,t
=
C
x,y,t
∪
C
x
+
1
,y,t
can be computed as follows:
pres
C
x
,y
,t
=
pres
C
x,y,t
+
pres
C
x
+
1
,y,t
−
C
x,y,t
.
crossX
Similarly, the values
C
x,y,t
.
crossY
and
C
x,y,t
.
crossT
can be used to compute
the presence in cells
C
x,y,t
∪
C
x,y
+
1
,t
and
C
x,y,t
∪
C
x,y,t
+
1
, respectively.
4.6.2 Querying the GeoPKDD TDW
We now use our query language
Q
agg
for querying the GeoPKDD TDW. As in
Section
4.4
, we assume a straightforward translation of the MultiDim schema
in Figure
4.5
into a snowflake schema. Notice that, because the TDW does not
contain moving object data, but only spatial data representing the partition of
the space, only SOLAP queries can be addressed to the TDW.
For example, Query
4.8
above reads in
Q
agg
:
SELECT d.name, sumPres
FROM District d
WHERE sumPres= SUM( SELECT a.presence
FROM AggTrajectories a, Cell c, Time t1, Time t2
WHERE a.cell=c.id AND contains(d.geometry,c.geometry)
AND a.startTime=t1.id AND a.endTime=t2.id
AND intersects(range(t1.dateTime,t2.dateTime),1/1/2010) )
