Normal Forms

for
Context-free Grammars

1
 Why Normal Form?

• Working with CFG, it is convenient to

have them in simplified form

• Easy one is Chomsky Normal Form

• Usually longer/expanded than initial

CFG

• Helpful for algorithms

2
 Chomsky Normal Form

A CFG is in Chomsky normal form, if

every rule is of the form:
A  BC or A a

Where a is any terminal and A, B, and C

are any variables- except that B and C
may not be the start variable.

3
 Chomsky Normal Form

Each production has form:

Type-II
Type 1
A  BC or A a

variable variable terminal

4
Examples:

𝑆→𝐴𝐶 𝑆→𝐴𝑆
Chomsky Not Chomsky
Normal Form Normal Form

5
 Conversion to Chomsky Normal Form

Example: S  ABa
Not Chomsky
A  aab Normal Form
B  Ac

We will convert it to Chomsky Normal Form

6
 The Algorithm – CFG to CNF
Step 1:
Make sure start symbol does not
Appear on right hand side.

Ɛ (Not allowed)
Step 2:

Unless S Ɛ (Allowed)
Remove Rules like A

Step 3:

B (Not allowed)
Get Rid of all unit rules
A
Reason: One non-terminal going to be another non-terminal is
Just an overhead in parse tree

Step 4:
Get rid of rules with more than 2 symbols on right hand side

Bcde (Not allowed)

A BCDE (Not allowed)
A
Step 5:
Make sure

a (only 1: Must be a Terminal symbol)

A BC (only 2: Must be Variables)
A
7
Example:

S  ASA | aB
A B|S
B  b|

Not Chomsky
Normal Form

8
 Conversion to Chomsky Normal Form

Example: S  ASA | aB
Not Chomsky
A B|S
Normal Form
B  b|

We will convert it to Chomsky Normal Form

9
 The Algorithm – CFG to CNF
Step 1:
Make sure start symbol does not
Appear on right hand side.

S0  S
S  ASA | aB
S  ASA | aB
A B|S
A B|S
B  b|
B  b|

10
 The Algorithm – CFG to CNF

Ɛ (Not allowed)
Step 2:

Unless S Ɛ (Allowed)
Remove Rules like A

S0  S
S  ASA | aB
A B|S
B  b|
S0  S
S  ASA | aB | a
A  B | S |
B b

11
 The Algorithm – CFG to CNF

Ɛ (Not allowed)
Step 2:
Remove Rule A

S0  S
S  ASA | aB | a
A  B | S |
B b

S0  S
S  ASA | aB | a | SA | AS | S
A B|S
B b

12
 The Algorithm – CFG to CNF
Step 3:

B (Not allowed)
Get Rid of all unit rules
A
Reason: One non-terminal going to be another non-terminal is
Just an overhead in parse tree

S0  S
S  ASA | aB | a | SA | AS | S
A B|S
B b

13
 The Algorithm – CFG to CNF
Step 3:

B (Not allowed)
Get Rid of all unit rules
A
Reason: One non-terminal going to be another non-terminal is
Just an overhead in parse tree

S0  S
S  ASA | aB | a | SA | AS | S
A B|S
B b
Remove S S
S0  S
S  ASA | aB | a | SA | AS
A B|S
B b
14
 The Algorithm – CFG to CNF
Step 3:

B (Not allowed)
Get Rid of all unit rules
A
Reason: One non-terminal going to be another non-terminal is
Just an overhead in parse tree

S0  S
S  ASA | aB | a | SA | AS
A B|S
B b
Remove S0 S
S 0  ASA | aB | a | SA | AS
S  ASA | aB | a | SA | AS
A B|S
B b
15
 The Algorithm – CFG to CNF
Step 3:

B (Not allowed)
Get Rid of all unit rules
A
Reason: One non-terminal going to be another non-terminal is
Just an overhead in parse tree

S 0  ASA | aB | a | SA | AS
S  ASA | aB | a | SA | AS
A B|S
B b
Remove A B
S 0  ASA | aB | a | SA | AS
S  ASA | aB | a | SA | AS
A b|S
B b
16
 The Algorithm – CFG to CNF
Step 3:

B (Not allowed)
Get Rid of all unit rules
A
Reason: One non-terminal going to be another non-terminal is
Just an overhead in parse tree

S 0  ASA | aB | a | SA | AS
S  ASA | aB | a | SA | AS
A b|S
B b
Remove A S
S 0  ASA | aB | a | SA | AS
S  ASA | aB | a | SA | AS
A  b | ASA | aB | a | SA | AS
B b
17
 The Algorithm – CFG to CNF
Step 4:
Get rid of rules with more than 2 symbols on right hand side

Bcde (Not allowed)

A BCDE (Not allowed)
A

e.g .
A  BCDE
Re placeWith
A  BA1
A1  CA2
A2  DE

18
 The Algorithm – CFG to CNF
Step 4:
Get rid of rules with more than 2 symbols on right hand side

Bcde (Not allowed)

A BCDE (Not allowed)
A

S 0  ASA | aB | a | SA | AS
S  ASA | aB | a | SA | AS
A  b | ASA | aB | a | SA | AS
B b
S 0  AA1 | aB | a | SA | AS
S  AA1 | aB | a | SA | AS
A  b | AA1 | aB | a | SA | AS
B b
A1  SA
19
 The Algorithm – CFG to CNF
Step 5:
Make sure

a (only 1: Must be a Terminal symbol)

A BC (only 2: Must be Variables)
A

Re place
A  bC
With
A  A1C
A1  b

20
 The Algorithm – CFG to CNF
Step 5:
Make sure

a (only 1: Must be a Terminal symbol)

A BC (only 2: Must be Variables)
A

S 0  AA1 | aB | a | SA | AS
S  AA1 | aB | a | SA | AS
A  b | AA1 | aB | a | SA | AS
B b S 0  AA1 | A2 B | a | SA | AS
A1  SA S  AA1 | A2 B | a | SA | AS
A  b | AA1 | A2 B | a | SA | AS
B b
A1  SA
A2  a
21
 The Algorithm – CFG to CNF
Step 5:
Make sure

a (only 1: Must be a Terminal symbol)

A BC (only 2: Must be Variables)
A

S 0  AA1 | A2 B | a | SA | AS
S  AA1 | A2 B | a | SA | AS
S  ASA | aB
A  b | AA1 | A2 B | a | SA | AS
A B|S
B b
B  b|
A1  SA
A2  a
Chomsky Not Chomsky
Normal Form Normal Form
22
 Observations

• Chomsky normal forms are good

for parsing and proving theorems

23
 Parsing Trees
• However, Chomsky Normal form has its
own limitations

• It is not most efficient for parsing trees

• This leads to emergences of other

standards/normal form

• One better for parsing is Greibach form

24