Designing Data Marts from XML and Relational Data Sources - Data Warehousing Design and Advanced Engineering Applications - page 67

Figure 5. Transition trees for the e-Ticket DTD

Relational Schema Generation
for the xML Source

name is nID : the concatenation of node
name n to the string ' ID' .

In this stage, the typed transition trees are trans-
formed into relational schemas based on the links
among the transition trees. This transformation is
conducted through the algorithm XML2R which
uses the following notation:

Algorithm XML2R( )

1. N R = the set of transition
trees reduced to roots, i.e. , with-
out any descendant.
2. N T = the set of nodes all of
whose children are leaves not an-
notated with #.3. For each node n ∈
N R do {
3.1

• Ref

n : the set of nodes that can be foreign
keys in the relation R n (constructed on
node n )
Build_R (n): a function that builds on node

R n = Build_R (n);

•

3.2 Add_attribute(n, R n );

n a relation called R n
ID (R

3.3 Mark_PK({n}, R n );

•

n ): a function that returns the primary
key of a relation R n
Add_attribute (a, R

}
4. For each node n ∈ N T do { // n
has only leave children not anno-
tated with #

•

n ): Adds attribute a as a
column to the relation R n
Mark_PK (pk, R

4.1 R n = Build_R (n);

•

n ): Marks the pk list of at-
tributes as the primary key of the relation
R n
Mark_FK (a, R

4.2 ID_found = False;

4.3 For each child a of n

do {
4.3.1

•

n ): Marks the a attribute as
a foreign key in the relation R n
ADD_ID (R

Add_attribute(a, R n );

4.3.2 If (arc(n, a) = ID)

•

n ): adds an artificial primary
key to the relation R n . The primary key

then {
4.3.2.1 Mark_