are associated with the entity's key identifier. Relationships are based on key
identifiers and can be between two unique entities, one unique identity to
several other entities, or between several entities.
A variety of techniques have been developed to sketch and make
schemes showing the data model. Usually these techniques follow the entity-
relationship conceptual understanding of a relational database. Entity-rela-
tionship diagrams, or E-R models, are often made using a graphical form
based on the Universal Modeling Language (UML), which offers a systematic
way of going from the diagram to conceptual and actual database descrip-
The capability of creating relations between data is extremely powerful
and useful. However, relations work only when data is stored using the same
format. For instance, returning to the address example from above, if the
entire address is stored as one database field, it will require additional pro-
cessing to relate this data with address data separated into multiple fields.
If the data can be related, the relation can be permanent or temporary.
Any database processing of a relation that produces a single, permanent new
table is called a “join”; otherwise it is just a relationship.
Data normalization is a process of assuring that a database can take best
advantage of relational database principles. If you normalize a database you
can not only improve its performance, but avoid some organizational and
logical errors that could diminish the quality of the database. Data normal-
ization of relationship database technology was first described by Edgar
Codd in the 1970s. The first level of data normalization requires that each
field contain only one value (e.g., only the house number, not the house
Entity-relationship (E-R) diagram.