A Method for Data Warehouse Design - Data Warehouse Systems: Design and Implementation - page 384

should eventually discover a collection of facts, measures, dimensions, and
hierarchies. The process includes several steps, as follows.

Define, Refine, and Prioritize Goals

The starting point in determining analysis needs is the consideration of the
business goals. Successful data warehouse projects assume that the goals of
the company are the same for everyone and that the entire company will
therefore be pursuing the same direction. Therefore, a clear specification of
goals is essential to guide user needs and convert them into data elements.
Since users at several different management levels participate in requirements
specification, analysis needs may be expressed by considering both general
and specific goals. The specific goals should be aligned with the general ones
to ensure a common direction of the overall development. The goal-gathering
process is conducted by means of interviews and brainstorming sessions,
among other ones. The list of goals should be analyzed to detect redundancies
and dependencies. For example, some goals could be combined because of
their similarity, discarded because of their inconsistency, or considered as
subgoals of other ones. This analysis may require additional interaction with
the users to establish the final list of goals.

Operationalize Goals

Once the goals have been defined and prioritized, we need to make them
concrete. Thus, for each goal identified in the previous step, a collection
of representative queries must be defined through interviews with the
users. These queries capture functional requirements , which define the
operations and activities that a system must be able to perform. Each user is
requested to provide, in natural language, a list of queries needed for her daily
task. Initially, the vocabulary can be unrestricted. However, certain terms
may have different meanings for different users. The analyst must identify and
disambiguate them. For example, a term like “the best customer” should be
expressed as “the customer with the highest total sales amount.” A document
is then produced, where for each goal there is a collection of queries, and
each query is associated with a user. The process continues with query
analysis and integration. Here, users review and consolidate the queries in the
document above to avoid misunderstandings or redundancies. The frequency
of the queries must also be estimated. Finally, a prioritization process is
carried out. Since we worked with different areas of the organization, we
must unify all requirements from these areas and define priorities between
them. A possible priority hierarchy can be areas queries of the

same user . Intuitively, the idea is that the requirement with the least priority
in an area prevails over the requirement with the highest priority in the

→ users →