Database Reference
In-Depth Information
Type 3 does not have a similarity relation defi ned in its domain. We name these attributes
as fuzzy attributes Type 4.
We defi ne a fuzzy attribute defi ned specialization just like an attribute defi ned special-
ization (Elmasri et al., 2000) where this attribute is a fuzzy attribute. It is represented with
an angled line joining the superclass with the circle. This line will be labeled with the name
of the fuzzy attribute Type n, preceded by the text “Tn:”. This defi nition is independent of
all constraints like fuzzy or crisp disjointed or overlapping specializations.
The following example shows two fuzzy attribute defi ned specialization (disjointed
and overlapping). In one specialization, each pair of subclasses has a fuzzy similarity degree
between them (Type 3). This property is useful to compare them and to search the more
important instances in some queries. In the other specialization, the similarity relation does
not exist (Type 4).
Example 8.
The conceptual model represented in Figure 9 expresses that in a real
estate agency, every landed property belongs to one subclass, which has its own attributes.
Thus, this is a total disjointed specialization (double line and “d” inside the circle). At-
tribute Kind is a fuzzy attribute Type 3, because if one person is looking for a chalet, for
example, then this customer is, possibly, interested in semi-detached houses because these
two types are similar. Thus, this is taken into account in order to show to our customer all
the relevant properties. In this sense, fuzzy queries are studied in Galindo et al. (1998, 1999)
and Galindo (1999). Observe that subclasses are not fuzzy, because every landed property
belongs only to one subclass.
Every landed property has an owner, which is a customer. Another kind of customer
is a claimant who is looking for a landed property. The overlapping specialization makes
it such that one customer may be owner and claimant at the same time. The fuzzy attribute
Type 4, Kind, makes it possible to store a possibility distribution about the subclasses in
order to express any fuzzy concept. In this example we are interested in measuring the
urgency of the customer. Thus, a customer with the value {0.4/Owner, 1/Claimantg} is a
customer who is looking for a landed property urgently and who is offering some property
Figure 9: Example 8. Two fuzzy attribute defi ned specializations
