Types of Normalization

Normalization usually occurs in phases where every phase is assigned its equivalent ‘Normal form’. As we progress upwards the phases, the data gets more orderly and hence less permissible to redundancy, and more consistent. The commonly used normal forms include:

• First Normal Form (1NF): In the 1NF stage, each column in a table is unique, with no repetition of groups of data. Here, each entry (or tuple) has a unique identifier known as a primary key.

• Second Normal Form (2NF): Building upon 1NF, at this stage, all non-key attributes are fully functionally dependent on the primary key. In other words, the non-key columns in the table should rely entirely on each candidate key.

• Third Normal Form (3NF): This stage takes care of transitive functional dependencies. In the 3NF stage, every non-principal column should be non-transitively dependent on each key within the table.

• Boyce-Codd Normal Form (BCNF): BCNF is the next level of 3NF that guarantees the validity of data dependencies. The dependencies of any attributes on non-key attributes are removed under the third level of normalization . For that reason, it ensures that each determinant be a candidate key and no dependent can fail to possess an independent attribute as its candidate key.

• Fourth Normal Form (4NF): 4NF follows that data redundancy is reduced to another level with the treatment of multi-valued facts. Simply put, the table is in normal form when it does not result in any update anomalies and when a table consists of multiple attributes, each is independent. In other words, it collapses the dependencies into single vs. multi-valued and eliminates the root of any data redundancy concerned with the multi-valued one.

The normalization concept for relational databases, developed by E.F. Codd, the inventor of the relational database model, is from the 1970s. Before Codd, the most common method of storing data was in large, cryptic, and unstructured files, generating plenty of redundancy and lack of consistency. When databases began to emerge, people noticed that stuffing data into them caused many duplications and anomalies to emerge, like insert, delete, and update anomalies. These anomalies could produce incorrect data reporting, which is harmful to any business. Normalization is a methodological method used in the design of databases to create a neat, structured, and structured table in which each table relates to just one subject or one-to-one correspondence.

The objective is to extensively reduce data redundancy and dependency. In essence, normalization was introduced and has continually been improved to rectify these specific aspects of data management. By organizing data in such a rigorous and stringent manner, normalization facilitates a significantly enhanced level of data integrity and enables more efficient data operations.