ai-week slot-and-filler structure

21
WEAK SLOT-AND-FILLER STRUCTURE 9.1 SEMANTIC NETS The main idea behind semantic nets is that the meaning of a concept comes from the ways in which it is connected . In a semantic net, information is represented as a set of nodes connected to each other by set of labled arcs, which repesent relationships among the nodes. Mammal Person Blue Brooklyn- Dodgers Pee-Wee-Reese Nose has-part isa instance team Uniform- color A Semantic network

Upload: jani442

Post on 28-Mar-2015

191 views

Category:

Documents


8 download

TRANSCRIPT

Page 1: AI-week slot-and-filler structure

WEAK SLOT-AND-FILLER STRUCTURE

9.1 SEMANTIC NETS

The main idea behind semantic nets is that the meaning of a concept comes from the ways in which it is connected . In a semantic net, information is represented as a set of nodes connected to each other by set of labled arcs, which repesent relationships among the nodes.

Mammal

Person

Blue Brooklyn-DodgersPee-Wee-Reese

Nosehas-part

isa

instance

teamUniform-color

A Semantic network

Page 2: AI-week slot-and-filler structure

9.1.1 Intersection search

9.1.2 Representing Nonbinary predicates

isa(person, Mammal)Instance(Pee-Wee-Reese, person)team(Pee-Wee-Reese,Brooklyn-Dodgers)uniform-color(Pee-Wee-Reese,Blue)

G23

Dodgers

Game

5-3Cubs

A semantic Net for an n-place predicate

isa

scoreVisiting team

Home-team

Page 3: AI-week slot-and-filler structure

EV7

Mary

Give

BK23john

A semantic Net representing sentence

instance

objectagent

beneficiary

BK23

instance

Page 4: AI-week slot-and-filler structure

H1

john

H2

height

Greater-than

Bill

height

john 72height

Making some important distinctions

H1

john

H2

height

Greater-than

Bill

height

H2

value

1.

3.

2.

Page 5: AI-week slot-and-filler structure

9.1.4 Partitioned Semantic Nets

Using partitioned semantics

d

Dogs

b

isa

assailant

Bite

isa

m

Mall-carrier

isa

victim

d

Dogs

b

isa

assailant

Bite

isa

m

Mall-carrier

isa

victim

S1

SA

g

GS

form

A

isa

a)

isa

b)

Page 6: AI-week slot-and-filler structure

d

Dogs

b

isa

assailant

Bite

isa

c

constables

isa

victim

S1

g

GS

form

A

isa

c)

Town-dog

SA

Page 7: AI-week slot-and-filler structure

d

Dogs

b

isa

assailant

Bite

isa

m

Mall-carrier

victim

gGS

S1

SA

A

isa

form

d)

Page 8: AI-week slot-and-filler structure

Feilderisa: ML-Baseball-Playercardinality: 376*bating-average: .262

Pee-Wee_Reeseinstance: Fielderheight: 5-10bats: Rightbating-average: .309team: Brooklyn-Dodgersuniform-color: Blue

ML-Baseball-Teamisa: Teamcardinality: 26*team-size: 24*manager:

Brooklyn-Dodgersinstance: ML-Baseball-Teamteam-size: 24manager: Leo-Durocherplayers: {pee-wee_Reese,…..}

Page 9: AI-week slot-and-filler structure

Representing the Class of All Teams as a Metaclass

Classinstance: classisa: class*cardinality:

Teaminstance: classisa: classcardinality: {the number of teams that exist}*team-size: {each team has a size}

ML-Baseball-Team:isa: Mammalinstance: classisa: Teamcardinality: 26{the number of baseball teams that exists}team_size: 24{default 24 players on a team}*manager:

Page 10: AI-week slot-and-filler structure

Brooklyn-Dodgers:instance: ML-Baseball-Teamisa: ML-Baseball-Playerteam-size: 24manager: Leo-Durocher*uniform-color: Blue

Pee-Wee-Reeseinstance: Brooklyn-Dodgersinstance: Fielderuniform-color: Bluebatting-average: .309

Page 11: AI-week slot-and-filler structure

Class(set of sets)

Team

ML-Baseball-Team

Pee-Wee-Reese

ML-Baseball-Player Brooklyn-Dodgers

Classes and Metaclasses

Page 12: AI-week slot-and-filler structure

9.2.3 Slots as Full-Fledged Objects

ML-Baseball-Player

Three-Finger-Brown

Pitcher National-leaguer

American-leaguer

FielderCatcher

instanceinstance

isaisaisaisaisa

Representing Relationships among Classes

Page 13: AI-week slot-and-filler structure

ML-Baseball-Player is covered by: {Pitcher,Catcher,Fielder}

{American-Leaguer,National-Leaguer}

Pitcherisa: ML-Baseball-Playermutually-disjount-with: {Catcher,Feilder}

Catcherisa: ML-Baseball-Player mutually-disjount-with: {Pitcher,Fielder}

Fielderisa: ML-Baseball-Playermutually-disjount-with: {pitcher,Catcher}

American-Leaguerisa: ML-Baseball-Playermutually-disjount-with: {National-Leaguer}

Page 14: AI-week slot-and-filler structure

National-Leaguerisa: ML-Baseball-Playermutually-disjount-with: {American-Leaguer}

Three-Finger-Browninstance: Pitcherinstance: National-leaguer

Page 15: AI-week slot-and-filler structure

Representing Slots as Frames,I

colormanager

my-manager

Uniform-color

bats

SlotClass(set of sets)

. .Pee-Wee-Reese Blue . .Smokey-The-Beat Brown . .

. .Pee-Wee-Reese Blue . .

Page 16: AI-week slot-and-filler structure

Slotisa: calssinstance: class*domain:*range:*range-constraint:*definition:*default:*transfers-through:*to-compute:*inverse:*single-valued:

Managerinstance: Slotdomain: ML-Baseball-Teamrange: Personrange-constraint: ƛx{baseball-experience x.manager}default:inverse: manager-ofsingle-vlaued: TRUE

Page 17: AI-week slot-and-filler structure

my-managerinstance: slotdomain: ML-Baseball-Playerrange: Personrange-constraint: ƛx(baseball-experience x.my-manager)to-complete: ƛx(x.team).managersingle-valued: TRUE

colorinstance: slotdomain: Physical-Objectrange: Color-Settransfers-through: top-level-part-ofvisual-salience: Highsingle-valued: FALSE

Uniform-colorinstance: Slotisa: colordomain: team-playerrange: color-setrange-constraint: not Pintvisual-saliencd: Highsingle-valued: FALSE

Page 18: AI-week slot-and-filler structure

batsinstance: Slotdomain: ML-Baseball-Playerrange: {Left,Right,Switch}to-compute: ƛx x.handedsingle-valued: TRUE

Associating Defaults with Slots

Bating-averageinstance: Slotdomain: ML-Baseball-Playerrange: Numberrange-constraint: ƛx(0<=x.range-constraint<=1)default: .252single-valued: TRUE

Fielder-batting-averageinstance: Slotisa: batting-averagedomain: Fielderrange: Numberrange-constraint: ƛx(0<=x.range-constraint<=1)single-valued: TRUE

Page 19: AI-week slot-and-filler structure

A Shorthand Notation for Slot-Range Specification

ML-Baseball-Playerbats: MUST BE {Left,Right,Switch}

Representing Slot-Values

Representing Slot-Values with Lambda Notation

9.2.4 Slot-Values as Objects

Johnheight: 72

Billheight:

Johnheight: 72;ƛx(x.height>Bill.height)

Billheight: ƛx(x.height<John.height)

Page 20: AI-week slot-and-filler structure

9.2.4 Inheritance Revisted

Tangled Hierarchies

Bird fly:yes

Ostrich fly:no Pet-Bird

Fifi fly:?

isa isa

instanceinstance

Quaker pacifist :no

Republican pacifist:false

DIck pacifist:?

instanceinstance

(a) (b)

Page 21: AI-week slot-and-filler structure

More Tangled Hierarchies

Bird fly:yes

Ostrich fly:no

Pet-Bird

isa isa

Fifi fly:?

instance

instanceWhite-Plumed Ostrich

Plumed- Ostrich

isa

isa

Quaker pacifist :no

Republican pacifist:false

DIck pacifist:?

instanceinstance

(b)

Conservative-Republican

isa

(a)