LOGICAL DEDUCTION IN AI

PROPOSITIONAL LOGIC

Arijit Mondal & Partha P Chakrabarti

Indian Institute of Technology Kharagpur
 Logic in Ancient Times
Indic Greek Today
Geometry, Calculations Thales, Pythagoras (Propositions Propositional
Nyaya, Vaisisekha and Geometry) Predicate
Theory of Argumentation Heraclitus, Parmenides (Logos) Higher Order
Sanskrit language with Binary- Plato (Logic beyond Geometry) Logic, Numbers &
Level arguments Aristotle (Syllogism, Syntax) Computation
Logical Argumentation: Chatustoki Stoics Psychology
Buddhist and Jain Philosophies Philosophy
Formal Systems Middle East
Vedanta Ancient Egypt, Babylon
Arab (Avisennian Logic)
China Inductive Logic
Confucious, Mozi,
Master Mo (Mohist School) Medieval Europe
Basic Formal Systems Post Aristotle
Buddhist Systems from India Precursor to First Order Logic
 First Few Examples
• If I am the President then I am well-known. I am the President. So I am well-
known
• If I am the President then I am well-known. I am not the President. So I am
not well-known.
• If Rajat is the President then Rajat is well-known. Rajat is the President. So
Rajat is well known.
• If Asha is elected VP then Rajat is chosen as G-Sec and Bharati is chosen
as Treasurer. Rajat is not chosen as G-Sec. Therefore Asha is not elected
VP.
• If Asha is elected VP then Rajat is chosen as G-Sec and Bharati is chosen
as Treasurer. Rajat is chosen as G-Sec. Therefore Asha is elected VP.
 Deduction Using Propositional Logic: Steps
Choice of Boolean Variables a, b, c, d, … which can take values true or
false.
Boolean Formulae developed using well defined connectors ~, Ʌ, V, →,
etc, whose meaning (semantics) is given by their truth tables.
Codification of Sentences of the argument into Boolean Formulae.
Developing the Deduction Process as obtaining truth of a Combined
Formula expressing the complete argument.
Determining the Truth or Validity of the formula and thereby proving or
disproving the argument and Analyzing its truth under various
Interpretations.
Deduction Using Propositional Logic: Example 1
Choice of Boolean Variables a, b, c, d, If I am the President then I am well-known. I am
… which can take values true or false. the President. So I am well-known
Boolean Formulae developed using well
defined connectors ~, Ʌ, V, →, etc,
Coding: Variables
whose meaning (semantics) is given by a: I am the President
their truth tables. b: I am well-known
Codification of Sentences of the
argument into Boolean Formulae.
Coding the sentences:
Developing the Deduction Process as F1: a → b
obtaining truth of a Combined Formula F2: a
expressing the complete argument.
G: b
Determining the Truth or Validity of the
formula and thereby proving or The final formula for deduction: (F1 Ʌ F2) → G,
disproving the argument and Analyzing that is:
its truth under various Interpretations. ((a → b) Ʌ a ) → b
Deduction Using Propositional Logic: Example 1
Boolean variables a, b, c, d, … which If I am the President then I am well-known. I am the President. So I am well-
known
can take values true or false.
Coding: Variables
Boolean formulae developed using well a: I am the President
defined connectors ~, Ʌ, V, →, etc, b: I am well-known
whose meaning (semantics) is given by Coding the sentences:
their truth tables. F1: a → b
F2: a
Codification of sentences of the
G: b
argument into Boolean Formulae.
The final formula for deduction: (F1 Ʌ F2) → G, that is: ((a → b) Ʌ a ) → b
Developing the Deduction Process as
obtaining truth of a combined formula
expressing the complete argument.
Determining the Truth or Validity of the
formula and thereby proving or
disproving the argument and Analyzing
its truth under various interpretations.
Deduction Using Propositional Logic: Example 2
Boolean variables a, b, c, d, … which If I am the President then I am well-known. I am not the President. So I am not well-
can take values true or false. known
Coding: Variables
Boolean formulae developed using
a: I am the President
well defined connectors ~, Ʌ, V, →,
b: I am well-known
etc, whose meaning (semantics) is
Coding the sentences:
given by their truth tables.
F1: a → b
Codification of sentences of the F2: ~a
argument into Boolean Formulae. G: ~b
Developing the Deduction Process The final formula for deduction: (F1 Ʌ F2) → G, that is: ((a → b) Ʌ ~a ) → ~b
as obtaining truth of a combined
formula expressing the complete
argument.
Determining the Truth or Validity of
the formula and thereby proving or
disproving the argument and
Analyzing its truth under various
interpretations.
Deduction Using Propositional Logic: Example 3
If I am the President then I am well-known. I am the President. So I am well- If Rajat is the President then Rajat is well-
known known. Rajat is the President. So Rajat is well
Coding: Variables known
a: I am the President Coding: Variables
b: I am well-known a: Rajat is the President
Coding the sentences: b: Rajat is well-known
F1: a → b Coding the sentences:
F2: a F1: a → b
G: b F2: a
The final formula for deduction: (F1 Ʌ F2) → G, that is: ((a → b) Ʌ a ) → b G: b
The final formula for deduction:
(F1 Ʌ F2) → G,
that is: ((a → b) Ʌ a ) → b
Deduction Using Propositional Logic: Example 4 & 5
If Asha is elected VP then Rajat is chosen as G-Sec If Asha is elected VP then Rajat is chosen as G-
and Bharati is chosen as Treasurer. Rajat is not Sec and Bharati is chosen as Treasurer. Rajat is
chosen as G-Sec. Therefore Asha is not elected VP. chosen as G-Sec. Therefore Asha is elected VP.
 More Examples
If Asha is elected VP then Rajat is chosen as G-Sec If Asha is elected VP then either Rajat is chosen as G-
or Bharati is chosen as Treasurer. Rajat is not Sec or Bharati is chosen as Treasurer but not both.
chosen as G-Sec. Therefore if Asha is elected as VP Rajat is not chosen as G-Sec. Therefore if Asha is
then Bharati is chosen as Treasurer elected as VP then Bharati is chosen as Treasurer
 Methods for Deduction in Propositional Logic
Interpretation of a Formula
Valid, non-valid, Satisfiable, Unsatisfiable
Decidable but NP-Hard
Truth Table Method
Faster Methods for validity checking:-
Tree Method
Data Structures: Binary Decision
Diagrams
Symbolic Method: Natural Deduction
Soundness and Completeness of a
Method
 Methods for Deduction in Propositional Logic
Interpretation of a Formula NATURAL DEDUCTION:
Valid, non-valid, Satisfiable, Unsatisfiable Modus Ponens: (a → b), a :- therefore b
Modus Tollens: (a → b), ~b :- therefore ~a
Decidable but NP-Hard
Hypothetical Syllogism: (a → b), (b → c):-
Truth Table Method therefore (a → c)
Faster Methods for validity checking:- Disjunctive Syllogism: (a V b), ~a:- therefore b
Tree Method Constructive Dilemma: (a → b) Ʌ (c → d), (a V
c) :- therefore (b V d)
Data Structures: Binary Decision
Destructive Dilemma: (a → b) Ʌ (c → d), (~b V
Diagrams ~d) :- therefore (~a V ~c)
Symbolic Method: Natural Deduction Simplification: a Ʌ b:- therefore a
Soundness and Completeness of a Conjunction: a, b:- therefore a Ʌ b
Method Addition: a :- therefore a V b
Natural Deduction is Sound and Complete
 Insufficiency of Propositional Logic
Wherever Mary goes, so does the lamb. Mary goes to school. So the
lamb goes to school.
No contractors are dependable. Some engineers are contractors.
Therefore some engineers are not dependable.
All dancers are graceful. Ayesha is a student. Ayesha is a dancer.
Therefore some student is graceful.
Every passenger is either in first class or second class. Each
passenger is in second class if and only if he or she is not wealthy.
Some passengers are wealthy. Not all passengers are wealthy.
Therefore some passengers are in second class.
Thank you