Construction and Maintenance of Heterogeneous Data Warehouses - Data Warehousing Design and Advanced Engineering Applications - page 197

Figure 3.

exists at least one path between each couple
of nodes. In this step we detect the related
connected sub-graphs of each graph G k . The
graph G k is then the union of its connected
sub-graphs. Let us note  k the set of these
sub-graphs:  k ={G ki / i∈T k } where T k is an
indexing set for  k

d) Generation of a view schema for each sub-

graph: We associate a view to each union of
sub-graphs realized. The views components
constitute the intersection of L' and all the
sub-components of the components associ-
ated to the nodes of these sub-graphs.

Example 10: By supplementing example9, two
view's schemas W 1 and W 2 are obtained. W 1 ={x 2 ,
x 3 , y 1 , y 8 , y 9 , y 10 } and W 2 ={x 5 , y 1 , y 8 , y 9 , y 10 }.
The fifth step of our approach ( Building the
Data warehouse ) gives the possibility to build the
views feeding the DW. For every view W to build,
the algorithm ConstrView() allows the integration
of various and heterogeneous data.
Let us note that:

Example 8: By supplementing example 7, G 1
is formed by two connected subgraphs G 11 et G 12
and G 2 contains only one connected subgrap G 21 =
G 2 (seeFigure 4). 1 ={G 11 , G 12 } et  2 ={G 21 }.

c) Unions of the related sub-graphs: it is real-

ized by taking each time only one connected
sub-graph from each graph G k . Let's have
T=∏ k∈D T k et d = |D|. for each u=(u 1 ,…,u d )
∈Τ we consider the graph R u formed by the
union of all the G ui (i=1,…,d). Since these
graphs do not have neither common nodes
nor common edges. The union of the con-
sidered graphs means their juxtaposition.

• A set of fictitious components (of CompNull

names) is created in order to ensure Union-
compatibility between the built queries if
SI w' = ∅.
The procedure Create_View(Q) consists in •

carrying out a selection query of the vari-
ous components appearing in Q. According
to the warehouse model this query can be
translated. Thus, the procedure depends
on the adopted modelling for the source
and the DBMS used. It also depends on

We denote by R the set of the obtained graphs:
R={R u / u ∈T}.
Example 9: By supplementing example 8, we
obtain the following set R:R={R 1 , R 2 } with R 1 =
G 11 ∪ G 21 and R 2 = G 12 ∪ G 21

Figure 4.