Sei sulla pagina 1di 34

Master of Science in Artificial Intelligence, 2014-2016

Knowledge
Representation
and Reasoning
University Politehnica of Bucharest
Department of Computer Science
Fall 2014
Adina Magda Florea

Lecture 1
Lecture outline
Course goals
Grading
Textbooks and readings
Syllabus
Why KR?
KR&R Challenges
What is KR&R?
Formal logic: why and how
Links for the young researcher
2

Course goals
Provide an overview of existing representational
frameworks developed within AI, their key
concepts and inference methods.
Acquiring skills in representing knowledge
Understanding the principles behind different
knowledge representation techniques
Being able to read and understand research
literature in the area of KR&R
Being able to complete a project in this research
area
3

Grading
Course grades
Mid-term exam
Final exam
Projects
Laboratory

20%
30%
30%
20%

Class participation

Bonus points

Requirements: min 7 lab attendances, min 50% of term


activity (mid-term ex, projects, lab)
Academic Honesty Policy
It will be considered an honor code violation to give or
use someone else's code or written answers, either for
the assignments or exam tests. If such a case occurs,
we will take action accordingly.

Textbooks and Readings


Textbooks

Artificial Intelligence: A Modern Approach (3rd Edition) by Stuart


Russell and Peter Norvig Prentice Hall, 2010
http://aima.cs.berkeley.edu/

Knowledge Representation and Reasoning by Ronald Brachman


and Hector Levesque, Morgan Kaufman, 2004
Artificial Intelligence: Foundations of Computational Agents by
David Poole, Alain Mackworth, Cambridge University Press, 2010

http://artint.info/index.html - available online


Computational Intelligence: a Logical Approach by David Poole,
Alain Mackworth, and Randy Goebel, Oxford University Press,
1998

Readings

Reading materials will be assigned to you.


You are expected to do the readings before the class
5

Syllabus
1. General knowledge representation issues
2. Logical agents Logical knowledge representation and
reasoning

First order predicate logic revisited


Nonmonotonic logics and reasoning
Modal logic, logics of knowledge and beliefs
Semantic networks and description logics, reasoning
services
Knowledge representation for the Semantic Web
6

Syllabus
3. Rule based agents
Rete: Efficient unification
The Soar model, universal subgoaling and chunking
Modern rule based systems
4. Probabilistic agents
Markov decision processes
Bayesian networks
Hidden Markov models
Dynamic Bayesian networks

Syllabus
5. Reasoning with actions
Planning
Introduction to robotics
6. Knowledge representation in learning

Inductive logic programming

7. Intelligence without representation and reasoning


vs. Strong AI

Calls Debate

Links for the young researcher

AI-MAS Links of interest


http://aimas.cs.pub.ro/links

Academic publishing
http://en.wikipedia.org/wiki/Academic_publishing

Writing a Scientific Paper


http://www.oup.com/us/samplechapters/0841234620/?view=usa

ISI Web of Knowledge


http://isiwebofknowledge.com/

Master Journal List


http://science.thomsonreuters.com/mjl/

Conference Proceedings Citation Index


http://wokinfo.com/products_tools/multidisciplinary/webofscience/cpci/

TED Ideas worth spreading


http://www.ted.com/
9

Lecture 1
Readings for Lecture 1:
http://plato.stanford.edu/entries/logic-ai/
Readings for Lecture 2
AIMA Chapter 7
http://aima.cs.berkeley.edu/newchap07.pdf

10

1. Why KR?
What is knowledge?
We understand by "knowledge" all kinds of
facts about the world.
Knowledge is necessary for intelligent
behavior (human beings, robots).
In this course we consider representation of
knowledge and how we can use it in making
intelligent artifacts(based on software,
hardware or both).
11

2. KR&R Challenges
Challenges of KR&R:

representation of commonsense knowledge


the ability of a knowledge-based system to
achieve computational efficiency for
different types of inferences

the ability to represent and manipulate


uncertain knowledge and information.

12

3. What is KR?
Randall Davis, Howard Shrobe, Peter Szolovits, MIT

A knowledge representation is most


fundamentally a surrogate, a substitute
for the thing itself, used to enable an entity
to determine consequences by reasoning
about the world.
It is a set of ontological commitments,
i.e., an answer to the question: In what
terms should I think about the world?
13

What is KR?
It is a theory of intelligent reasoning
comprising:
the representation's fundamental
conception of intelligent reasoning;
the set of inferences the representation
suppor

It is a medium of human
expression, i.e., a language in which
we say things about the world.
14

What is KR?
If A represents B, then A stands for B and
is usually more easily accessible than B.
Symbolic representations
Non-symbolic representations

15

4. What is Reasoning?
Reasoning is the use of symbolic
representations of some statements in
order to derive new ones.
Inference a form of reasoning
Use of inferences (rules of inference)

16

5. Models of KRR

Symbolic logic and ATP


Probabilistic
Rules
Structured

17

6. Formal logic
Formal logic is the field of study of entailment
relations, formal languages, truth conditions,
semantics, and inference.
All propositions/statements are represented as
formulae which have a semantics according to
the logic in question.
Logical system = Formal language +
semantics
Formal logics gives us a framework to discuss
different kinds of reasoning.

18

6.1 Logical consequence (entailment)


Proof centered approach to logical
consequence: the validity of a reasoning
process (argument) amounts to there
being a proof of the conclusions from the
premises.

19

Logical consequence (entailment)


Model centered approach to logical
consequence
Models are abstract mathematical structures that
provide possible interpretations for each of the
objects in a formal language.
Given a model for a language - define what it is
for a sentence in that language to be true
(according to that model) or not.

20

6.2 Model centered approach


Interpretation of a formula
Model of a formula
Entailment or logical consequence
A formula F is a logical consequence of a set of
formulas P1,Pn iff F is true in all interpretations in
which P1,Pn are true.
P1, Pn |= L F
T Formula F is a logical consequence of a set of
formulas P1,Pn iff P1,Pn F is valid.
T Formula F is a logical consequence of a set of
formulas P1,Pn iff P1 Pn ~F is inconsistent.
21

6.3 Proof centered approach


Theorem, deduction
Formal system
Inference rule

S =< A , F , A , >
R
R

R F n F , y = y1 ,..., y n x, x, y i F , i = 1, n

Premise set

= {y1 , ... , yn }

E1 = E 0 U {x| y E 0n , y x}
n 1

Consequence of

E0 = A

E 2 = E1 U {x| y E1n , y x}
n 1

E i ( i 0)
22

Proof centered approach


If E 0 = A then x E i is deductible from
|S x

Theorems - the elements of Ei if E 0 = A ( = )


x Ei
Demonstration

| R x, x is povable
23

6.4 Properties of logical systems


Important properties of logical systems:
Consistency - no theorem of the system contradicts
another.
Soundness - the system's rules of proof will never
allow a false inference from a true premise. If a system
is sound and its axioms are true then its theorems are also guaranteed to
be true.

Completeness - there are no true sentences in the


system that cannot, at least in principle, be proved
in the system.
Some logical systems do not have all three properties. Kurt Godel's
incompleteness theorems show that no standard formal system of
arithmetic can be consistent and complete.
24

Properties of logical systems


A logical system L is complete iff
|= L implies |
(i.e., all valid formulas are provable)
A logical system L is sound iff
| implies |= L
(i.e., no invalid formula is provable)
FOPL
Second order logics
25

7. Logic based representations


2 possible aims

to make the system function according to the logic


to specify and validate the design

Conceptualization of the world / problem


Syntax - wffs
Semantics - significance, model
Model - the domain interpretation for which a formula is true
Model - linear or structured
M |=S - " is true or satisfied in component S of the structure M"

Model theory
Generate new wffs that are necessarily true, given that the old wffs
are true - entailment
KB |=L
Proof theory
Derive new wffs based on axioms and inference rules
KB |-i
26

PrL, FOPL

Linear model

Extend PrL, PL
Sentential logic
of beliefs
Uses beliefs atoms BA()
Index PL with agents

Situation calculus
Adds states, actions
Symbol level
Knowledge level

Modal logic
Modal operators
Structured models

Description Logics
Subsumption relationships

Logics of knowledge
and belief
Modal operators B and K

Temporal logic
Modal operators for time
Linear time
Branching time
CTL logic
Branching time
and action

Dynamic logic
Modal operators
for actions

BDI logic
Adds agents, B, D, I

27

knowledge

propositional first-order

Paul is a man

man(Paul)

Bill is a man

man(Bill)

men are mortal

(x) (man(x)
mortal(x))
First order logic
c

knowledge

first-order

second-order

smaller is
transitive

( x) (( y) (( z)
((<(x,y) <(y,z)
<(x,z)))))
( x) (( y) (( z)
((part-of(x,y)
part-of(y,z)
part-of(x,z)))))

transitive(<)

part-of is
transitive
R is transitive iff

not expressible

whenever R(x,y) and


R(y,z) hold, R(x,z)
holds too

(see however pseudosecond order)

Higher order logic

transitive(part-of)

( R) ((transitive(R)
( x) (( y) (( z)
((R(x,y) R(y,z)
R(x,z)))))))

28

29

A logical puzzle
Someone who lives in Dreadbury Mansion killed Aunt
Agatha.
Agatha, the butler, and Charles live in Dreadbury
Mansion, and are the only people who live therein.
A killer always hates his victim, and is never richer than
his victim.
Charles hates no one that Aunt Agatha hates.
Agatha hates everyone except the butler.
The butler hates everyone not richer than Aunt Agatha.
The butler hates everyone Aunt Agatha hates.
No one hates everyone.
Agatha is not the butler.
Who killed Aunt Agatha?
30

% % Who killed Aunt Agatha puzzle


% % Used predicates :
% % lives(X) : X lives in Dreadbury Mansion
% killed(X, Y) : X killed Y
% richer(X, Y) : X is richer than Y
% hates(X, Y) : X hates Y
% Rem : X = Y is equivalent to the predicate equal(X, Y)
% X != Y is equivalent to ~equal(X, Y)
% equal(X, Y) is a predefined predicate
% If the equal predicate wasn't defined, we could use
% somes axioms to have the same behaviour.
31

% Axioms for equality :


% % premise{ ![X] : equal(X, X) }
% premise{ ![X,Y,Z] : (equal(X,Y) =>
% ((lives(X) <=> lives(Y)) &
% (killed(X,Z) <=> killed(Y,Z)) &
% (killed(Z,X) <=> killed(Z,Y)) &
% (richer(X,Z) <=> richer(Y,Z)) &
% (richer(Z,X) <=> richer(Z,Y)) &
% (hates(X,Z) <=> hates(Y,Z)) &
% (hates(Z,X) <=> hates(Z,Y)))) }

32

% Someone who lives in Dreadbury Mansion killed Aunt Agatha.


premise{ ?[X] : (lives(X) & killed(X, agatha)) }
% Agatha, the butler, and Charles live in Dreadbury Mansion,
premise{ lives(agatha) } premise{ lives(butler) } premise{ lives(charles) }
% and are the only people who live therein.
premise{ ![X] : (lives(X) => (X = agatha | X = butler | X = charles)) }
% A killer always hates his victim,
premise{ ![X,Y] : (killed(X, Y) => hates(X, Y)) }
% and is never richer than his victim.
premise{ ![X,Y] : (killed(X, Y) => ~richer(X, Y)) }
% Charles hates no one that Aunt Agatha hates.
premise{ ![X] : (hates(agatha, X) => ~hates(charles, X)) }
33

% Agatha hates everyone except the butler.


premise{ ![X] : ((X != butler) => hates(agatha, X)) }
% The butler hates everyone not richer than Aunt Agatha.
premise{ ![X] : (~richer(X, agatha) => hates(butler, X)) }
% The butler hates everyone Aunt Agatha hates.
premise{ ![X] : (hates(agatha, X) => hates(butler, X)) }
% No one hates everyone.
premise{ ![X] : ?[Y] : ~hates(X, Y) }
% Agatha is not the butler.
premise{ agatha != butler }
% Therefore : ?X killed Agatha.
conclusion{ killed(?X, agatha) }

34

Potrebbero piacerti anche