Introduction to Knowledge Representation

1
Introduction to Knowledge Representation

• Marti Hearst
• SIMS 202 Information Organization and Retrieval
• Lecture 6, Sept 10, 1998

2
Today

• What is a symbol?
• Semantics the meanings of symbols
• Creating Ontologies
• Objects, Properties, and Relations
• References
• Chapter 1 of Introduction to Knowledge Systems
  by Mark Stefik.
• Chapter 8 of Artificial Intelligence, A Modern
  Approach by Stuart Russell and Peter Norvig

3
What is a symbol?

• From Merriam-Websters Collegiate
• Something that stands for or suggests something
  else.
• An arbitrary conventional sign used in writing or
  printing to represent
• operations
• quantities
• elements
• relations
• qualities
• What is meant by sign and represent?

4
Recognizing Symbols

• What is/are this/these symbol(s)?
• Two overlapping squares?
• Eight horizontal and eight vertical lines?

5
Designation and Representation

• What does it mean to represent something?
• Identify the following
• symbols
• context
• observer
• Representation is the association of symbols with
  conceptual objects or ideas in a given context.
• The observer sets up a correspondence between the
  symbols and the meanings.

6
Representation with Symbols
Kailin threw the ball to Juno.
did-action
Juno
thrown-to
Kailin
throw event
object- thrown
a ball
thrower
7
Symbols and Language

• Abstract concepts are difficult to express in a
  computer.
• Combinations of abstract concepts are even more
  difficult to express
• time
• shades of meaning
• social and psychological concepts
• causal relationships

8
Symbols and Language
The Dog.
9
Symbols and Language
The Dog.
The dog cavorts.
The dog cavorted.
The picture doesnt really show the manner or
tense.
10
Symbols and Language
The man.
The man walks.
11
Symbols and Language
The man walks the cavorting dog.
So far, we can sort of show the meaning in
pictures.
12
Symbols and Language
As the man walks the cavorting dog,
thoughts arrive unbidden of the previous spring,
so unlike this one, in which walking was marching
and dogs were baleful sentinals outside unjust
halls.
What is the relation between the symbols and the
meaning?
13
Symbols and Language

• Language only hints at meaning.
• Most meaning of text lies within our minds and
  common understanding.
• How much is that doggy in the window?
• how much social system of barter and trade (not
  the size of the dog)
• doggy implies childlike, plaintive, probably
  cannot do the purchasing on their own
• in the window implies behind a store window,
  not really inside a window, requires notion of
  window shopping

14
Setting up Correspondences between Symbols and
Meaning

• Consider made-up languages
• Codes used by espionage agents
• Pope means a particular piece of microfilm
• Denver indicates a particular mailbox
• People remember the gist instead of the actual
  words used.
• This implies the actual words used are not very
  salient what matters is the meaning.

15
Recognizing Symbols

• The marks that constitute a symbol depend on the
  conventions for recognizing the symbol.
• A recognizer typically has an associated alphabet
  or set of symbols
• Token an individual instance of a symbol
• Type a class of symbols
• Examples?

16
The Role of Context

• The concept associated with the symbol 21 means
  different things in different contexts.
• Examples?
• The question Is there any salt?
• Asked of a waiter at a restaurant.
• Asked of an environmental scientist at work.

17
Semantics The Meaning of Symbols

• Semantics versus Syntax
• Meaning versus Representation
• What a persons name is versus who they are.
• A rose by any other name...
• What the computer program looks like versus
  what it actually does.

18
Semantics

• Semantics assigning meanings to symbols and
  expressions.
• Usually involves defining
• objects
• properties of objects
• relations between objects
• More detailed versions include (among others)
• events
• time
• places
• measurements (quantities)

19
Ontology

• From Merriam-Websters Collegiate
• A branch of metaphysics concerned with the nature
  and relations of being.
• A particular theory about the nature of being or
  the kinds of existence.
• More prosaically
• A carving up of the worlds meanings.
• Determine what things exist, but not how they
  inter-relate.
• Related terms
• taxonomy, dictionary, category structure

20
Knowledge Engineering Steps

• Decide what to talk about
• Decide on a vocabulary of predicates, functions,
  and constants
• Encode general knowledge about the domain
• Artificial Intelligence vs Cataloging
• AI goal allow computer programs to reason about
  the objects and relations
• Cataloging organize the objects and relations
  for use by humans
• AI is more ambitious and more difficult
• We arent covering the reasoning part here.

21
Try some examples

• Lets define
• Types of Objects
• Types of Properties of Objects
• Types of Relations between Objects

22
Attributes vs. Objects

• How do we make this distinction?
• Say we are clothing manufacturers.
• Fur is a class of objects
• Animal is an attribute of this class
• Say we are naturalists.
• Animal is a class of objects
• Fur is an attribute of this class

23
Garment Maker Ontology

• Define the objects Indicate what types of
  attributes are used to define the objects
  (attributesproperties)
• Object Class
• Garment
• Attribute Types
• ISA
• Material
• Color
• Garment_Type
• Object Class
• Fur
• Attribute Types
• ISA
• Animal
• Color
• Texture

24
Garment Maker Ontology

• Attributes have lists of legal values
• Object Class Garment
• ISA Object
• Material fur, cotton, wool
• Color red, black, brown, white, blue
• Garment_Type coat, stole, hat
• Object Class Fur
• ISA Material
• Animal fox, rabbit, sable
• Color red, black, white
• Texture silky, thick, coarse

25
Garment Maker Ontology

• Show the assignments of values to attributes for
  one particular instance of an object
• Object Class Garment
• ISA Object
• Material fur, cotton, wool
• Color red, black, brown, white, blue
• Garment_Type coat, stole, hat
• Object Class Fur
• ISA Material
• Animal fox, rabbit, sable
• Color red, black, white
• Texture silky, thick, coarse

object
ISA
garment
Material
G_type
fur
coat
Color
Animal
Texture
26
Garment Maker Ontology

• Usually only one value is allowed for an ISA
  attribute
• In this example,
• The value of the color attribute for Garment is
  determined by the color attribute for the
  garments Material attribute
• This is called inheritance

27
Garment Makers vs. Naturalists

• A difference between a class definition and an
  attribute value
• Class Fur
• ISA material
• Animal fox, rabbit, sable
• Color red, black, white
• Texture silky, thick, coarse
• Garment_type coat, stole, hat
• Class Animal
• ISA mammal
• Outer_Covering fur, skin, scales
• Number_of_limbs 4, 6, 8
• Circulatory_System cold_blooded, hot_blooded

28
Nesting Attributes and Classes

• Class Garment
• Material
• Class Fur
• Animal fox, rabbit, sable
• Color red, black, white
• Texture silky, thick, coarse
• Class Cotton
• Color red, blue, white, brown, black
• Thread_Count 100, 200
• Garment_type stole, coat, hat, t-shirt
• Attributes often must be nested
• Alternative two subclasses of Garment

29
Next Week

• Semantic Nets
• Facets vs. Hierarchies
• Lexical Semantics
• Word Associations