Information Technology Reference
In-Depth Information
(03/2<((
3HUVRQ
(PSOR\HHIODJ
$OXPQXVIDOJ
6WXGHQWIODJ
R
VHW
VHW
VHW
(PSOR\HH
$OXPQXV
6WXGHQW
(PSOR\HH
$OXPQXV
6WXGHQW
1HWZRUNVFKHPD
FRUUHVSRQGLQJ((5PRGHO
Fig. 3.11 Map isa relationship to overlap generalization
%DQN
3HUVRQ
&RPSDQ\
%DQN
3HUVRQ
&RPSDQ\
VHW
VHW
VHW
X
2ZQHU
2ZQHU
FRUUHVSRQGLQJ((5PRGHO
1HWZRUNVFKHPD
Fig. 3.12 Map isa relationships to categorization in EER model
relationships with one owner and more than one member record type. However,
user input is needed to confirm these semantics. Figure 3.11 is an example with Em-
ployee-flag, Alumnus-flag, and Student-flag being used to indicate the membership
of the subclass entities (Employee, Alumnus, Student). An employee can be both a
student and a person. The difference between disjoint and overlap generalization is
that the former needs only one predicate field while the latter needs one predicate
field for each subclass entity.
A categorization is derived by mapping isa relationships and their record types
to superclass/subclass entities relationships such that a set of class entities (mapped
from a set of owner record types) can be united to form the superclass entity of a
subclass entity (mapped from a member record type). Again, the knowledge acqui-
sition system should be able to detect a potential categorization by locating isa re-
lationships with more than one owner and one member record type. However, user
input is needed to confirm such a semantic. Figure 3.12 is an example.
Substep 7—Derive entity keys and other constraints.
There are three forms of identifiers. They can be described as follows:
• Fullyinternallyidentified—Therecordkeyuniquelyidentifiestherecord.For
example, in a loan system, records can be identified as in Fig. 3.13 . Here Loan#
and Collateral# are unique in the whole loan system.
 
Search WWH ::




Custom Search