50%(2)1 out of 2 people found this document helpful
This preview shows page 9 - 10 out of 19 pages.
ous styles in different publications and organizations. In Figure 2-15a, the potentially multiple prerequisites of a course (shown as a multivalued attribute in the Attribute cell) are also courses (and a course may be a prerequisite for many other courses). Thus, prerequisite could be viewed as a bill-of-materials structure (shown in the Relationship & Entity cell) between courses, not a multivalued attribute of COURSE. Representing prerequisites via a bill-of-materials structure also means that finding the prerequisites of a course and finding the courses for which a course is prerequisite both deal with relationships between entity types. When a prerequisite is a multivalued attribute of COURSE, finding the courses for which a course is a prerequisite means looking for a specific value for a prerequisite across all COURSE instances. As was shown in Figure 2-13a, such a situation could also be modeled as a unary relationship among instances of the COURSE entity type. In Visio, this specific situation requires creating the equiva-lent of an associative entity (see the Relationship & Entity cell in Figure 2-15a; Visio does not use the rectangle with rounded corners symbol). By creating the associative entity, it is now easy to add characteristics to the relationship, such as a minimum grade required. Also note that Visio shows the identifier (in this case compound) with a PK stereotype symbol and boldface on the component attribute names, signifying these are required attributes.In Figure 2-15b, employees potentially have multiple skills (shown in the Attribute cell), but skill could be viewed instead as an entity type (shown in the Relationship & Entity cell as the equivalent of an associative entity) about which the organization wants to maintain data (the unique code to identify each skill, a descriptive title, and the type of skill, for example, technical or managerial). An employee has skills, which are not viewed as attributes, but rather as instances of a related entity type. In the cases of Figures 2-15a and 2-15b, representing the data as a multivalued attribute rather than via a relationship with another entity type may, in the view of some people, simplify the diagram. On the other hand, the right-hand drawings in these figures are closer to the way the database would be represented in a standard relational database management system, the most popular type of DBMS in use today. Although we are not concerned with implementation during conceptual data modeling, there is some logic for keep-ing the conceptual and logical data models similar. Further, as we will see in the next example, there are times when an attribute, whether simple, composite, or multivalued, should be in a separate entity.So, when shouldan attribute be linked to an entity type via a relationship? The answer is when the attribute is the identifier or some other characteristic of an entity type in the data model and multiple entity instances need to share these same attri-butes. Figure 2-15c represents an example of this rule. In this example, EMPLOYEE has a composite attribute of Department. Because Department is a concept of the busi-