Knowledge Representation

1
Knowledge Representation

• Issues and Brief History
• Semantic Networks
• Conceptual Dependency Graphs
• Scripts and Frames
• Summary

2
Issues and Brief History
Initially knowledge representation was mostly
based on the use of logic.
Languages with sound and complete inference
rules.
Philosophers and Mathematicians
3
Alternative Research
An alternative line of research came from the
effort of psychologists and linguists. Goal
Characterize the nature of human
understanding. Associationist theories
Define the meaning of an
object using a network of
associations.
4
Alternative Research
Through experience, we associate concepts with
other concepts.
Cold White Snowman Slippery Ice
Snow
5
Knowledge Representation

• Issues and Brief History
• Semantic Networks
• Conceptual Dependency Graphs
• Scripts and Frames
• Summary

6
Semantic Networks

• Collins and Quillian (1969) modeled human
  information
• using semantic networks.
• Humans
• Associate concepts
• Organize knowledge hierarchically

7
Figure 6.1 Semantic network developed by Collins
and Quillian in their research on human
information storage and response times (Harmon
and King 1985).
8
Semantic Networks

• Advantages of Semantic Networks
• Information can be stored at the highest level
  of abstraction
• If we need to specify a concept in more detail,
  we do it
• after inheriting the features of the more
  general concept.
• Example describe a particular type of bird.
• Knowledge is based on association of concepts.
• Can be represented as graphs (nodes ? facts or
  concepts,
• links ? relations).

9
Semantic Networks History

• Much work has been done in natural language
• understanding.
• First implementations were used in machine
• translation (1960s).
• Quillian (1967) created a program that
  illustrates how
• semantic networks work in the context of a
  dictionary.
• Given two words it would search the graphs in a
  
• breadth-first fashion until a common
  intersection is found.

10
Figure 6.3 Three planes representing three
definitions of the word plant (Quillian 1967).
11
Figure 6.4 Intersection path between cry and
comfort (Quillian 1967).
12
Knowledge Representation

• Issues and Brief History
• Semantic Networks
• Conceptual Dependency Graphs
• Scripts and Frames
• Summary

13
Conceptual Dependency Graphs

• Semantic networks represent relationships poorly
  
• (they unable to cope with complex domains).
• Roger Schank (1974) proposed conceptual
  dependency graphs
• to capture the semantic structure of
  natural language.
• Relations are captured through primitives.
• Four primitive conceptualizations are as
  follows
• Actions
• Objects
• Modifiers of actions
• Modifiers of objects

14
Conceptual dependency theory of four primitive
conceptualizations
15
Figure 6.6 Conceptual dependencies (Schank and
Rieger 1974).
16
Figure 6.9 Conceptual dependency representing
John ate the egg(Schank and Rieger 1974).
Figure 6.10 Conceptual dependency representation
of the sentence John prevented Mary from giving
a book to Bill (Schank and Rieger 1974).
17
Advantages of Conceptual Dependency Graphs

• Reduce ambiguity with a formal theory of
  semantics.
• The representation captures semantics in a
  canonical way
• two sentences may be written differently
  but if the meaning
• is the same, they would end up represented
  with the same
• graph.

18
Knowledge Representation

• Issues and Brief History
• Semantic Networks
• Conceptual Dependency Graphs
• Scripts and Frames
• Summary

19
Scripts and Frames
Scripts and frames are ways to capture
prototypical situations. If we read about
baseball, or politics, or sports, we tend to
immediately bring to our mind a stereotyped
situation describing those scenarios.
20
Example

• As an example, think of what you would expect in
• a restaurant
• You are met at the entrance by someone or there
  is a sign
• indicating you can find a seat.
• Either a menu is available or it is presented by
  the waiter.
• There are routines for ordering food, eating,
  paying, leaving.

21
(No Transcript)
22
(No Transcript)
23
Figure 6.12 Part of a frame description of a
hotel room. Specialization indicates a pointer
to a superclass.
24
Knowledge Representation

• Issues and Brief History
• Semantic Networks
• Conceptual Dependency Graphs
• Scripts and Frames
• Summary

25
Summary

• One way of storing knowledge is by placing
  importance
• on concepts and relations.
• Semantic networks capture knowledge as
  associations
• of concepts in a hierarchical manner.
• Conceptual dependency graphs capture the
  semantic
• structure of natural language.
• Scripts and frames represent stereotyped
  situations.