Database and Expert Systems Integration - Information Systems Reengineering, Integration and Normalization - page 309

sibly transform, the data of the lower tier into knowledge usable by the integrated
system. The upper tier combines and enhances the knowledge of the existing system
with additional knowledge to create an integrated expert database system
Tier 1: Existing Systems
The existing systems contain data to be reused in the new/integrated EDS. Only the
data required for the operation of the integrated system is extracted. This data is
brought into a consistent state through the coupling classes of tier 2.
Tier 2: Coupling Classes
Coupling classes describe the information in existing systems. A coupling class
provides the interface between the extension (or reengineered upper) layer and the
existing systems. The uniformity of this interface layer insulates the upper layers
from changes in the lower layers and can be used to bring information together so
that data representing the same entities or attributes are consistent.
An attribute in a coupling class is derived from the values of the entities stored in
the underlying systems. The derivation is a simple one-to-one mapping.
The coupling classes provide information from existing knowledge repositories,
and additional information can also be stored by the integrated system. The infor-
mation from the existing systems is only extracted on demand, as it would be un-
wise to copy information out of these repositories to store in the integrated systems
without endangering the consistency of information across the organization.
Tier 3: Integrated Classes
The third layer combines the components of the coupling classes with additional
classes (and objects) to create an integrated system. To form an integrated system,
name conflict and semantic conflict problems need to be solved. Since the system
has a unified structure (i.e., a higher level synthesis model), the name conflict prob-
lem can be easily solved by using the synonym index. The synonym index creates a
relationship between two different attributes with the same values.
To solve the semantic conflict problem between different attributes, additional
classes must be appended into the integrated system. For instance, the value “vacan-
cy” in the employment attribute of the Employee relation in an existing relational
database indicates that the employee is available for assignment to a new project. In
an existing ES, the same information is represented using an attribute availability
with values “yes” or “no”. To resolve this conflict, an additional class must be cre-
ated to enable the availability attribute of the existing ES to derive its value from
the Employee database coupling class. The new additional class must involve the
following method:
IF employment = “vacancy”
THEN availability = “yes”
ELSE availability = “no”