DATA MODELING BASICS - Database Design and Development: An Essential Guide for IT Professionals - page 159

Integrating Partial Data Models

In this approach, you do not wait to apply data modeling techniques until all the
requirements are merged. You create a mini-model or partial data model from a
few related user views. How do you form these sets of related user views? Usually,
user views of a department or a natural user group can be combined to form a
related group. This approach for creating the consolidated data model by integrat-
ing partial data models generally consists of the following major tasks. Amend and
adapt this list to suit your database project.

1. Combine the data elements of related user views (usually those of a depart-
ment or user group).
2. Eliminate duplicate data elements.
3. Separate out objects and their characteristics for the combined set of
elements.
4. Establish relationships among objects.
5. Create a local or mini data model for the objects and relationships for the first
combination.
6. Repeat tasks 1 through 5 until local data models are created for all
combinations.
7. Integrate all the local models, two at a time.
8. Indicate how an instance of each object will be uniquely identified.
9. Review the final, global data model created from the partial model for
completeness.

Figure 5-16 describes the approach of integrating partial or local data models for
the banking example. Observe how each task is shown in the figure and how the
tasks work together to produce the final data model.

Enhancement and Validation

As soon as you produce the global or complete data model with either of the two
methods, your data model is almost ready to complete the logical design. Before
you use the data model, you have to review the data model to ensure completeness.
This is the time to enhance the data model by removing discrepancies and to vali-
date the model.
Generally, enhancement of the data model includes the following tasks:

1. Reconcile naming conflicts . If two or more different names are used to
refer to the same data element, then these names are synonyms. Review
the data model and reconcile synonyms. If the same name is used to refer to
two or more data elements, then the name is a homonym. Reconcile

homonyms.
2. Resolve conflicts between data element types . If a data element in the model
refers to an object in one place and to a characteristic of another object in a
different place, then there is conflict between the two types of data elements.
Resolve such discrepancies.