Normalization in SQL ((1NF, 2NF, 3NF, BCNF, 4NF, 5NF)

Normalization:
Normalization is step-by-step process of reducing complexity of an entity by distributing the attributes to different entities. Following six normal rules are used sequentially on each entity to normalize them and the attribute that does not adhere to the rule is taken out of the entity.

  First Normal Form (1NF)

o   Every attribute must have value. It should not be empty. In RDBMS a column without value has NULL value that can be manipulated or accesses like a normal value.

o   Every attribute must have atomic value that cannot be shorten further. For example: Name is non-atomic attribute as it can be broken into First Name, Middle Name and Surname.

  Second Normal Form (2NF)

o   An entity must be in First Normal Form

o   Entity must have a Primary Key or Composite Primary Key

o   Every attribute must be fully and functionally dependent upon Primary Key

  Third Normal Form (3NF)

o   An entity must be Second Normal Form.

o   There must not be any dependency among non-key attributes (other that Primary Key)

  Boyce-Codd Normal Form (BCNF)

o   An entity must be in Third Normal Form.

o   If the entity has a Composite Primary Key & Alternate Primary Key with one or more attribute common to both, in such case entity must be broken into three entities.

 Fourth Normal Form (4NF)

o   An entity must be in Boyce-Codd Normal Form.

o   If an attribute is based on Value List must be taken out as a separate entity.

  Fifth Normal Form (5NF)

o   An entity must be in Fourth Normal Form.

o   If an attribute is multi-valued attribute then it must be taken out as a separate entity.

 

Normalization in SQL

Database Design Example:

This example takes you into step-by-step approach for designing a small database for Resume Management that is based on Resume Attributes.

Step-1: (Gathering Attributes for Resume Management):

            Personal:

Name, Father Name, Address, Date of Birth, Contact, Marital Status, Gender, Nationality, Religion

            Education:

Degree/Certificate Name, Pass Year, Institute Name, Marks Obtained (%), Rating

            Experience:

Employer Name, Designation, Join Date, Resign Date, Reason of Resignation, Roles

            Project:

Project Name, Client Name, Start Date, Duration (Weeks), Team Size, Role, Description

Step-2 (Defining Entities using Noun Approach):

Entities below are derived from the list of attributes using a simple approach of Noun. Under this approach look for all attributes and whatever words are Noun pick them as entity and list them.

Step-3 (Relation among entities):

            Now define the relations among entities as per the following rules:

§  One-to-One Relation (Link directly both the entities)

§  One-to-Many Relation (Link directly both the entities)

§  Many-to-Many Relation (Take a third entity and relate both the entities via third entity)

Normalization in SQL

 

Conventions in E-R Diagram:

§  Normal color rectangle boxes are Real Entities (Person, Religion, Father, etc.).

§  Inverse color rectangle boxes are Intermediate Entities to break Many-to-Many relation based Entities (PersonAddress, PersonContact, PersonDesignation, etc.).

§  Arrow indicates One-to-Many relation from Parent towards Child entity (One Religion Many Person, One Nationality Many Person, One Person Many Degree, etc.).

Rules for defining relations between Entities:

Rule #1:

If two entities are based on One-to-One relation then keep them as separate entities are merge them into single entity, preferred method is keep them as separate entity. In above E-R Diagram there are no two entities based on One-to-One relation.

Rule #2:

If two entities are based on One-to-Many relation then always keep them as separate entities and relate them from One (Parent Entity) towards Many (Child Entity). In above E-R Diagram Religion & Person, Nationality & Person are examples of One-to-Many relations.

Rule #3:

If two entities are based on Many-to-Many relation then introduce a third entity to relate both the entities. Third entity becomes Child Entity for both the Parent Entities. In above E-R Diagram Person & Degree (PersonDegree), Person & Address (PersonAddress), Person & Contact (PersonContact) are examples of Many-to-Many relations.

Rule #4:

If one entity (A) has relation with two entities (B & C) and both have their child entity (BC) then relate A with BC. In above E-R Diagram, Institute Entity has relation with Person Entity & Degree Entity and they both have their child entity PersonDegree so Institute Entity is related to PersonDegree entity.

Rule #5:

Rule #4 can be extended further If one entity (A) has relation with three Entities (B, C & D) and they have their Child Entities then we must relate A with the lowest level Child Entity. In above E-R Diagram, Project Entity has relation with Person Entity, Designation Entity & Employer Entity and they have their child entity EmployerPersonDesignation at the lowest level of relation so Project Entity is related to it. Since relation between Project and EmployerPersonDesignation is Many-to-Many so intermediate entity ProjectEmployerPersonDesignation is required.

Step-4 (Distribution of Attributes):

Tips:

§  Distribute the attributes among entities using common knowledge, if you think a particular attribute should be under an entity, keep there. At the time of Normalization process those attributes that do not belong to entity will be out from there.

§  If an entity remains with one attribute still keep entity separate as in future there might be some more attributes that may not visible right now.

Step-5 (Normalization of Entities):

First Normal Form:

 

            Since Name is non-atomic attribute so it is broken into three atomic attributes.

Second Normal Form:

            Person:                       

            Person ID (PK)          

            First Name                 

            Middle Name             

            Last Name                  

            Marital Status             

            Gender                                   

            Date Of Birth             

            Tips:

§  Check each attribute with following question:

<attribute> IS FOR?

If the answer is Entity then keep the attribute there else take it out. 

For example:

So the Person table is in Second Normal Form since all of its attributes are fully dependent upon Primary Key.

Third Normal Form:

            Person:                       

            Person ID (PK)          

            First Name                 

            Middle Name             

            Last Name                  

            Marital Status             

            Gender                                   

            Date Of Birth             

Person entity is already in Third Normal Form since none of the attributes have dependency among them.

Tips:

§  To check the dependency among attributes use the following approach:

Take one attribute and check that any other attributes’ Existence or Value is controlled by this attribute. For example: Marital Status decides whether Spouse Name will exists or not.

Boyce-Codd Normal Form:

            Person:                       

            Person ID (PK)          

            First Name                 

            Middle Name             

            Last Name                  

            Marital Status             

            Gender                                   

            Date Of Birth             

Person entity is already in Boyce-Codd Normal Form since it does not have Composite Primary Key and Alternate Primary Key.

Tips:

§  If you are following the rule of Noun based entity selection then BCNF case may never come into existence since chances of Composite PK and Alternate Composite PK are rare.

Fourth Normal Form:

            Person:                       

            Person ID (PK)          

            First Name                 

            Middle Name             

            Last Name                  

Date Of Birth             

Marital Status and Gender attributes are removed since they are based on List of Values (Married & Unmarried, Male & Female).

Tips:

§  Marital Status and Gender will be separate entities and attached to Person Entity in the form of One-to-Many relation (One gender many persons, One marital status many persons)

§  Add two more entities to earlier E-R Diagram and relate them with Person Entity.

 

Fifth Normal Form:

            Person:                       

            Person ID (PK)          

            First Name                 

            Middle Name             

            Last Name                  

Date Of Birth             

Person entity is already in Fifth Normal Form since none of the attribute keeps multiple values. For example Phone No attribute may keeps one or more phone numbers.