Migrating Data to an Integrated Database - High-Performance Web Databases: Design, Development, and Deployment - page 602

Exhibit 42-1. Indicating the correspondence between a source data
element and a target data element.
indicate how a data element (i.e., RegistrationNumber) in the source file
maps to a data element (i.e., RegNumber) in the target file.
Frequently, the names of two corresponding data items in the source
and target files are similar, so the database administrator has no trouble
identifying the correspondence. However, at times, data elements with the
same name or nearly the same name may have very different meanings. For
example, “date” may have several different meanings, including start date,
end date, date the data was entered, publication date, date of issue, or date
something becomes active. Database administrators must understand how
the data will be used to determine precisely if two data elements really
mean the same thing.
Sometimes, data elements with different names mean the same thing, as
in title and job, level and class, rank and position, and location and
address. Again, database administrators must understand how data will be
used to determine whether data items with different names are equivalent.
Different Abstraction Levels
Two similar data items may represent similar information at different
levels of abstraction. For example, the data item BirthDate is more specific
than BirthYear because BirthDate includes the month and day of birth. In
another example, the data element Author is binary and indicates only
whether a person has authored one or more topics. However, the element
NumberBooksAuthored is much more specific; it indicates that a person is
an author and the number of topics authored.
It is always possible to define a mapping from the specific to the more gen-
eral data element. Exhibit 2 illustrates the mappings from BirthDate to Birth-
Year and from NumberBooksAuthored to Author. However, additional infor-
mation is required to map a general data item to a more specific data item.
Retaining as much specificity as possible is recommended, so information
from the source database is not lost as it is migrated to the target database.