Database Normalization

Status: draft · Confidence: medium (0.635) · Basis: verified_sources

Quality notes: generic_source_homepage, no_verified_sources, partial_source_verification

## TL;DR

Database normalization reduces data redundancy and anomalies by organizing data into well-structured tables. Normal forms: 1NF (atomic values), 2NF (no partial dependencies), 3NF (no transitive dependencies), BCNF (every determinant is a candidate key). Most databases target 3NF — enough to eliminate most anomalies while remaining practical.

## Core Explanation

1NF: each cell holds single value, each row unique. 2NF: non-key attributes depend on the whole primary key (not part). 3NF: non-key attributes depend on nothing but the primary key. Denormalization: intentionally introduce redundancy for read performance (common in analytics/OLAP). Trade-off: normalization = write-optimized, denormalization = read-optimized.

## Further Reading

- [Database Systems: The Complete Book (Garcia-Molina, Ullman, Widom)](undefined)

## Related Articles

- [Batch Normalization](../../ai/batch-normalization.md)
- [Learned Database Systems: AI-Driven Query Optimization, Learned Indexes, and Cardinality Estimation](../../ai/learned-database-systems.md)
- [Text-to-SQL: Natural Language Database Querying with Large Language Models](../../ai/text-to-sql.md)