Database and Expert System Technology - Information Systems Reengineering, Integration and Normalization - page 31

Loan
Contracts
1

9

14

Loan
Drawdown
2

Loan
Interest
3

Loan
Repayment
6

Loan
Balence
8

10

11 7 15

Fixed
Rate
4

Index
Rate
5

13

12

Fig. 2.1 Hierarchical database of a loan system

• Eachrelationshipis1:nanditistotal.Thatis,ifR i is the parent of R j in the
hierarchy, then for every record occurrence of R j there is exactly one R i record
connected to it.

To construct a hierarchical model, it is natural to build an entity-relationship (ER)
model and map it to a hierarchical model because an ER model carries more seman-
tics. Once an ER model is built, if relationships are all binary, we can map a 1:n or
1:1 relationship from A to B as a binary tree. To map a m:n relationship from A to B,
we can use virtual record types (pointer to actual records) which are distinguished
by an identifier (ID) field in a physical address as shown in Fig. 2.2 (McElreath
1981 ).

2.1.1 Hierarchical Data Definition Language

Two types of structures are used to implement the inverted tree structure of a hi-
erarchical model, namely data definition and data occurrence trees. The role of a
data definition tree is to describe the data types and their relationships. For example
Fig. 2.1 shows seven data types, in a parent (the top one)-child (the bottom one) re-
lationship with respect to each other. Data occurrence tree represents the actual data
in the database. Figure 2.1 shows fifteen data occurrences in hierarchical sequence,
the default read sequence in hierarchical model.