Database Reference
In-Depth Information
The Enhanced Entity-Relationship Model (EER) (Connolly & Begg, 1998; Elmasri et
al., 2000; Hammer & McLeod, 1981) is an extension of the Entity-Relationship Model (ER)
(Chen, 1976). This study is based on the EER model published in Connolly et al. (1998),
and Elmasri et al. (1985, 2000), which is one of the most modern, versatile and complete
versions.
Fuzzy databases (Galindo, 1999; Medina et al., 1994; Petry, 1996) have also been widely
studied with scant attention being paid to the problem of conceptual modeling. At the same
time, the extension of the ER model for the treatment of fuzzy data (with vagueness) has
been studied in some publications (Chaudhry et al., 1994, 1999; Chen & Kerre, 1998; Ma
et al., 2001; Zvieli & Chen, 1986), but none of them refer to the possibility of expressing
constraints fl exibly by using the tools offered by fuzzy sets theory. Besides, these approaches
are not exhaustive in other senses. Perhaps the most exhaustive fuzzy modeling tool is the
FuzzyEER model (Galindo et al., 2001b, 2003, 2004; Urrutia et al., 2002a, 2002b, 2003)
and in this chapter we expose some of its advantages.
Zvieli and Chen (1986) allow fuzzy attributes in entities and relationships and they
introduced three levels of fuzziness in the ER model:
1.
At the fi rst level, entity sets, relationships and attribute sets may be fuzzy; namely,
they have a membership degree to the model. For example, the fuzzy entity Radio
may have a 0.7 importance degree as an integrating part of a car.
2.
The second level is related to the fuzzy occurrences of entities and relationships.
For example, an entity Young Employees must be fuzzy, because its instances, its
employees, belong to the entity with different membership degrees.
3.
The third level concerns the fuzzy values of attributes of special entities and relation-
ships. For example, attribute Quality of a basketball player may be fuzzy.
The fi rst level may be useful, but at the end we must decide whether such an entity,
relationship or attribute will appear or will not appear in the implementation. The second
level is useful too, but it is important to consider different degree meanings (membership
degree, importance degree, fulfi llment degree...). A list of authors using different meanings
is included in Galindo et al. (2001a). The third level is useful, but it is similar to writing the
data type of some attributes, because fuzzy values belong to fuzzy data types.
Chaudhry et al. (1994; 1999) propose a method for designing Fuzzy Relational Da-
tabases (FRDB) following the extension of the ER model of Zvieli et al. (1986), taking
special interest in converting crisp databases into fuzzy ones. The way to do so is to defi ne
linguistic labels as fuzzy sets and to obtain the membership degree to each of them of the
crisp value existing in the database.
In 1998, Chen et al. introduced the fuzzy extension of several major EER concepts
(superclass, subclass, generalization, specialization, category and subclass with multiple
superclasses) without including graphical representations. The proposal of Vert et al.
(2000) is based on the notation used by Oracle and uses fuzzy sets theory to treat data sets
as a collection of fuzzy objects, applying the result to the area of Geospatial Information
Systems (GIS).
Finally, Ma et al. (2001) work with the three levels of Zvieli and Chen (1986) and they
introduce a Fuzzy Extended Entity-Relationship (FEER) model to cope with imperfect as
well as complex objects in the real world at a conceptual level. However, their defi nitions
(of generalization, specialization, category and aggregation) impose very restrictive condi-