---

University of California, Davis

Application Development Methodology

---

Development Technique Concepts

Normalization

The decomposition of a data structure, to remove any implicit dependencies between objects within the structure.

Normalization is a staged procedure that reduces any given collection of attributes to several, more basic collections called "normalized relations." A fully normalized relation is a group of attributes in which the existence of values for the non-identifying attributes is entirely dependent upon the existence of values for the identifying attributes of the group.

Normalization is used within correctness checking to help decide whether an attribute has been assigned to the correct entity type, and whether further entity types are needed in the model.

Normal Form

The results of the stages of normalization are attribute groups in some normal form: First Normal Form (1NF), Second Normal Form (2NF), Third Normal Form (3NF), Fourth Normal Form (4NF) and Fifth Normal Form (5NF).

Each normal form serves to remove ambiguities from the database. In an unnormalized collection of attributes, the values can be organized in repeating groups (e.g., multivalued attributes) with many dependencies between them. In a 5NF collection of attributes, you will have separated out all multivalued attributes and eliminated all but the most direct dependencies and provided the most flexible access to unambiguous data.

First Normal Form

1NF achieves the separating of multivalued attributes by constructing a distinct separate collection of entity types to contain the multiple values. In 1NF, no entity type has repeating groups.

Second Normal Form

2NF achieves the separation of entity types that are not dependent on the complete identifier of the group. In 2NF, no entity types are only partially dependent on their identifiers.

Third Normal Form

3NF achieves the separation of entity types that are dependent on attributes other than the complete identifier. In 3NF, no entity types are indirectly dependent on their identifiers. A well designed entity relationship diagram is a third normal form schema.

Fourth Normal Form

4NF achieves the separation of entity types that are conditionally dependent on other attributes. In 4NF, every entity type is directly dependent on its identifier under all circumstances. Using entity subtypes in an entity relationship diagram places the schema into the fourth normal form.

Fifth Normal Form

5NF achieves the separation of large composite identifiers into independent tables, each with only a pair of identifying attributes. This can achieve a reduction in the level of redundancy among the total set of attribute values, but because it adds little to the meaningfulness of the model, it is generally not employed.