Information Technology Reference
In-Depth Information
2.2 Constraints and Dependencies in Conceptual Schemas
When designing a conceptual schema, it is indeed important to define a set of
constraints that will guarantee that once the subsequent database is implemented
and operational, any change made to its content by authorised users will main-
tain its consistency. Typically, inserting, modifying or deleting values from the
database should not result into data anomalies or unnecessary redundancies. For
this purpose, let us introduce the
relational model
of a database according to the
First normal form
(1NF), which is a database model based on first-order pred-
icate logic, first formulated and proposed by Codd [3]. In the relational model,
all the data is represented through
relation
s(alsoknownas
tables
). A relation
is composed of
attribute
s (a.k.a. as
columns
),eachofwhichisdefinedona
do-
main
, which is a given set of values. A
tuple
(a.k.a.
row
) contains all the data
of a single instance, that is, a value for each attribute of the relation. Relations
and attributes of the relational model will be used to model entity types and
attributes in the GER model
1
, as illustrated in Figure 1.
The transition between conceptual and relational models is facilitated in the
RAINBOW approach since the complex conceptual schemas obtained from the
form-based interfaces are recursively transformed until they reduce to simple
schemas including flat entity types (with elementary attributes) and binary re-
lationship types. Coping with these relationship types in the 1NF relational
modeling requires a little trick; the roles are represented through
role attributes
,
so that they appear as attributes for which the value belongs to the possible
tuples of the entity type associated through the binary relationship type. In
addition, when no value is provided in a form for a given attribute, a default
“empty” value (noted
∅
) is encoded, so that null values are avoided.
Fig. 1.
The representation of a Customer using the GER and relational model
Among the many
constraints
usually found in database schema, we have se-
lected the following ones.
Domains of values
restrict the possible values of given
attributes, for instance using domain types, sets or ranges of (un)authorised
values, rule-based formulas for the values, and so on. For instance, a
Customer
Number
may be restricted to integer values, and the
Gender
limited to the set
of values
{
Female,Male
}
.
Cardinality constraints
define the minimal (typically
1
The Generic Entity-Relationship (GER) model is a wide-spectrum model used to
describe database schemas in various abstraction levels and paradigms [4].
