Business Objects

The data architect/analyst produces a list of the business objects represented on the Context Model and then must make his or her own investigation of the business by touring the business area, collecting as many source documents and exhibits as they are able to, directly from the business user. This allows the data analyst to verify the context model with the user. Validate the stuff they use, how it's organized, and what the language means.

THis exercise also provides an opportunity to find out if the context model is missing anything. If so, add in the new business objects.

This also provides an opportunity to get the user's phone number and set the stage for follow-up questions. These documents and exhibits are data-oriented and are the primary inputs to the database design process. The database should be designed from data and information requirements.

Good data analysis sources are screen snapshots from existing systems, fill-in-the-blanks documents, computer-generated reports and spreadsheets. These exhibits show "what data", and also "how used". These are the business user's view of their data.

Poor data analysis sources are file and table definitions from existing application databases. These exhibits show "what data" and "how organized". These are a technician's view of the data. Most problems associated with existing systems are caused by the existing database design. These resources should be examined in subsequent iterations of the design process, but avoided in the first pass.

The worst data analysis sources are narrative documents, such as user and operations manuals. With respect to their authors, most of these are poorly written and the language is imprecise. These exhibits show neither "what data", "how used" or "how organized". These documents are process-oriented, not data-oriented.

All of the exhibits selected by the data analyst as input to the database design process should be catalogued by type of exhibit, name of exhibit, source of exhibit, use, creator and users.

The data analyst prepares for the Normalization process by organizing the input documents and exhibits in a pile with the simplest exhibits on the top and the more complex exhibits toward the bottom of the pile. By normalizing the easy ones first, the process begins more quickly and by the time the complex exhibits are analyzed, the data analyst has developed an understanding of the application that will help them with their analysis.