Normalization — Introduction

What is Normalization?

Normalization is the process of organizing a relational database to reduce data redundancy and improve data integrity by decomposing large, poorly structured tables into smaller, well-structured tables.

It was formalized by E.F. Codd as part of the relational model theory.

Why Normalization?

Consider this unnormalized table:

Problems (Anomalies)

1. Insertion Anomaly Cannot add a new course (CS103) without an enrolled student — all columns must have values.

2. Update Anomaly If Dr.R changes, we must update every row where CS101 appears. Miss one → inconsistency.

3. Deletion Anomaly If Bob drops MATH201, we lose all information about that course entirely.

Goals of Normalization

1. Eliminate redundancy — each fact stored in one place only
2. Eliminate update anomalies — changing one fact changes it everywhere
3. Eliminate insertion/deletion anomalies — can add/remove data independently
4. Ensure data integrity — constraints enforce correctness automatically

Normalization Process (Overview)

Each higher normal form assumes the previous one is already satisfied.

Decomposition

Normalization works by decomposing a table into two or more smaller tables such that:

1. Lossless Join — The original table can be reconstructed by joining the decomposed tables.
2. Dependency Preservation — All functional dependencies are maintained in the decomposed tables (so constraints can be enforced).

Normalized Version of the Example

Now:

• Adding a new course doesn't require a student
• Changing a professor affects only the Course table
• Dropping an enrollment doesn't lose course information