Part II Methods of AI

1
Part IIMethods of AI

• Chapter 3
• Knowledge and Reasoning

2
Part II Methods of AI
Chapter 3 Knowledge Representation and Reasoning
3.1 Summary of Logic and Reasoning
3.2 Reasoning Deduction Systems
3.3 Rulebased Reasoning
3.4 Knowledge Representation General Issues
3.5 Knowledge Representation Semantic Nets
3.6 Knowledge Representation Description Logics
3
3.4 Knowledge Representation
General Issues
4
Three Historical Paradigmsin Artificial
Intelligence
1. General Problem Solver versus Knowledge based
System 2. Knowledge Representation Procedural
versus declarative KR 3. R. Brooks
Representationless Representation
5
Good old fashioned AI (GOFAI-1) General
Principles for Symbol Manipulation in AI
I use intelligent methods so I think
Paradigm One had not to cheat by giving the
problem-solver the solution in any sense. (Pat
Hayes 1974 )
But Humans dont use only general methods for
problem solving, but also domain specific
knowledge
And there are inherently knowledge intensive
very specific tasks
6
Good old fashioned AI (GOFAI-2) General
Principles for Symbol Manipulation in AI
I use knowledge - so I think!

• Representation of Knowledge
• Effective methods to apply this knowledge

TOPICS
7
New fashioned AI (NeFAI-3) Intelligence
without Representation ?
Let the world be its own representation and I
think

• Neuronal Nets
• Rodney Brooks
• Patty Maes
• etc

• Symbolic versus Subsymbolic Representation
• Subsymbolic Repn - Methods Subsumption
  Architecture
• 
  Neural Networks
• 
  . . . and others.

TOPICS
8
3.5 Knowledge Representation
Semantic Networks
9
Semantic Networks
10
Simple and Naive Semantic Nets
11
Naive Semantic Networks to get it started
is-A

• I.

ROBIN
BIRD
is-A
is-A
II.
ROBIN
BIRD
CLYDE
is-A
is-A
III.
ROBIN
BIRD
CLYDE
has-part
WINGS
12
Semantic Nets an unstructured Example
PROBLEM-1 Semantics of the SN?
PROBLEM-2 Structure of the SNs ?
13
Another Example Winstons Arch
14
Semantic Networks 1 Schönfinkeln

• Binary Relations versus Relations with Many
  Argument
• Translation a la Schönfinkel
  (and other logicians)
• Example

Own(Owner, Ownee, Start, End)
Owner(own-1, ... ) Ownee(own-1, ...
) Start(own-1, ... ) End(own-1, ... )
Translates into
15
Semantic Networks 1 An Example for n-ary Relation
is-A
is-A
ROBIN
BIRD
CLYDE
is-A
ownee
NEST_1
NEST
OWN_1
is-A
start-time
Spring
TIME
is-A
end-time
Fall
is-A
SITUATION
OWNERSHIP
16
Token-Type Distinction in SNs
is-A
is-A
ROBIN
BIRD
CLYDE
owns
is-A
NEST_1
NEST
Token Nest_1 Type Nest
Nest_1 is of type Nest
17
Semantic Networks
Generic Concept An Arch

• Generic Concept
• Individual Concept
• Individual Object

PARIS
An Arc de Triomphe
18
Inferences with Semantic Nets
CAN-FLY
BIRD
property
is-A
color
YELLOW
CANARY
is-A
owns
SYLVESTER
TWEETY
? CAN-FLY (CANARY)
? CAN-FLY (TWEETY)
? SYLVESTER owns Something that can fly
? TWEETY is YELLOW
? SYLVESTER owns a CANARY
? SYLVESTER owns a BIRD
19
Contradictions in Semantic Nets
color
CANARY YELLOW SAM GREEN
is-A
color

• Inconsistencies in SNs ? Pro and Con!
• Operations on SNs ? Non-monotonic Reasoning

20
Semantic Networks

• Just a Graphical Representation of the Predicate
  Calculus ?

is-A
TWEETY
BIRD
Is-A (TWEETY, BIRD)
21
Naïve Network Theory Semantics
NETWORK

PK1
Is-A(ROBIN, BIRD)
LISP
(is-A ROBIN BIRD)

• Pidgin English Semantics?
• Infix (ROBIN IS-A BIRD)
• English ROBIN is a bird

22
Example Problems of Semantics
has-part
BIRD
WINGS
is-A
is-A
ROBIN
CLYDE
is-A
studied-by
ENDANGEREDSPECIES
NATURALISTS
has-part(BIRD,WINGS) ? has-part(ROBIN,WINGS)
Property(ENDANGERED SPECIES) ? Property(CLYDE)
Do the naturalists study the special bird CLYDE ?
23
Nice Properties of Semantic Networks

• Inheritance of Attributes
• Propagation of Parts
• Concept Centered Representation
• Semantic Distance
• Procedural Semantics of SNs
• BUT How to get a Declarative (Tarski) or
  Denotational Semantics of SNs?

24
Semantics

• Woods (1975) Whats in a Link?
• Question What is the Semantics of Semantic
  Networks?
• The Intuition of the Reader?
  
• The LISP-Programs operating
  on the SN?
• compare the early work on
  the semantics of
• programming
  languages
• Levesque Mylopoulos (1977)
• Procedural Semantics
• Cercone Schubert (1979)
• Translation into First Order Predicate Logic

25
Distinction of Representational and Conceptual
Levels in Semantic Networks