In designing relational databases, there are five normal forms: 1NF, 2NF, 3NF, 4NF, and 5NF.

Normalization is a method to correct potential problems or anomalies in relational databases.

The FD: A -> B implies the FD: B -> A

( A -> B and A -> C ) implies B -> C

The FD: A -> B means that each value for A is associated with exactly one value of B.

The FD: SID -> ( L , F , GPA ) is in the Student table as defined below: Student ( SID , L , F , GPA )

We can always determine a FD by just carefully examining the contents/records of a table or relation.

A , which is a column or a set of columns, is a candidate key for table R if all columns in R are functionally dependent on A, and no subset of A can functionally determine all columns in R.

A table can have many primary keys since a table may have many candidate keys.

An un-normalized table is not in 1NF.

In a relational DBMS, we can create a table that is not in 1NF.

Table W ( ON, D , TD, C , S , Q ) is not in 2NF if it has the following 3 FDs: FD1: ON -> D FD2: TD -> C FD3: ON, TD -> ( D , C , S , Q )

Given table R ( A , B , D , E ) with the following FDs: FD1: A -> B , D , E FD2: D -> E R is not in 3NF.

Given table R (SID,CID) , there are always 2 trivial FDs: SID -> CID and CID -> SID

Table W ( N, D , T , C , S ) is in 2NF if it has the following 2 FDs: FD1: N -> D FD2: N, T -> ( D , C , S )

A table T is in 4NF iff T is in 3NF and each determinant is a candidate key.

Valid multi-valued dependencies must NOT come in pairs.

The table Take ( SID , CID, grade) is in 4NF even though there are 2 trivial MDs and one FD (in the Take table) as follows: MD1: SID ->-> CID MD2: CID ->-> SID FD1: SID, CID -> grade

Table W ( N, D , T, S, Q ) is not in 2NF if it only has the following FD: FD1: N, T -> ( D, S , Q )

The Professor table (as defined below) is in 1NF, 2NF, and 3NF, but not in 4NF. Professor ( ProfID , StudentID , ClubID )

Given table R ( A , D , E ) with the following FDs: FD1: A -> D , E FD2: D -> E R is in 3NF.

A table should have a primary key, but a table may have many candidate keys.

A , which is a column or a set of columns, is a candidate key for table R if and only if all columns in R are functionally dependent on A.

The FD: SID -> GPA is in the Student table as defined below: Student ( SID , L , F , GPA )

A table T is in 3NF iff T is in 2NF and each determinant is a candidate key.

X -> -> Y if and only if Y -> -> X

Table W ( ON, D, TD, DC, SD, Q ) does not have any problems at all if it has only the following 3 FDs: FD1: ON -> D FD2: TD -> DC FD3: ON, TD -> ( D, DC , SD , Q )

A table that contains a repeating group (or multiple entries for a column in a single record) is not in 1NF.

Normally, a DBA would design a database with tables whose FDs are documented clearly in users’ manuals/policies.

We can create an un-normalized table in Access or any relational DBMS.

An un-normalized table can be in 1NF sometimes.

The FD: ( L , F ) -> SID is NOT in the Student table as defined below: Student ( SID , L , F , GPA )

The FD: F -> SID is in the Student table as defined below: Student ( SID , L , F , G )

( A -> B and B -> C ) implies A -> C

Normalization process may not guarantee to make all tables free of all anomalies or problems.

Table in 1NF is better than the table in 2NF, which is better than the table in 3NF or 4NF.