View
220
Download
0
Tags:
Embed Size (px)
Citation preview
Copyright © 2002 Cycorp
A Bundle of Assertions
• Think of a microtheory (mt) as a set of assertions.
• Each microtheory bundles assertions based on– a shared set of assumptions on which the truth of the assertions depends, or– a shared topic (world geography, brain tumors, pro football), or– a shared source: (CIA World Fact Book 1997, FM101-5, USA Today)
the Cyc KB, as a sea of assertions
Copyright © 2002 Cycorp
Avoiding Inconsistencies
• The assertions within a microtheory must be mutually consistent– no monotonic contradictions allowed within a single microtheory
• Assertions in different microtheories may be inconsistent
the Cyc KB, as a sea of assertions
in MT1: tables, etc., are solidin MT2: tables are mostly space
in MT1: Mandela is an elder statesmanin MT2: Mandela is President of South Africain MT3: Mandela is a political prisoner
Copyright © 2002 Cycorp
Every Assertion is in a Microtheory
• Every assertion falls within at least one microtheory
• Currently, every microtheory is a reified (named) term, such as #$HumanActivitiesMt or #$OrganizationMt
• Mts are one way of indexing all the assertions in Cyc
Copyright © 2002 Cycorp
Better/faster/more scalable knowledge base building Better/faster/more scalable inferencing, too.
• To focus development of the Cyc knowledge base• To enable shorter and simpler assertions
Mandela is president vs. Mandela is president throughout 1995 in South Africa
Tables are solidvs. At granularity usually considered by humans, tables are solid
•To cope with global inconsistency in the KB, inevitable at this scale• Each mt is locally consistent (content in unrelated mts is not visible)• Good for handling divergence (different points of view, scientific theories, changes over time)
Why Have Microtheories?
Copyright © 2002 Cycorp
Better/faster/more scalable knowledge base building Better/faster/more scalable inferencing, too.
• To focus development of the Cyc knowledge base• To enable shorter and simpler assertions
Mandela is president vs. Mandela is president throughout 1995 in South Africa
Tables are solidvs. At granularity usually considered by humans, tables are
solid
•To cope with global inconsistency in the KB, inevitable at this scale• Each mt is locally consistent (content in unrelated mts is not visible)• Good for handling divergence (different points of view, scientific theories, changes over time)
Why Have microtheories? (cont.)
Copyright © 2002 Cycorp
•#$VocabularyMicrotheory -- each instance contains definitions of general concepts used in a knowledge domain •(e.g., #$TransportationVocabMt, #$ComputerSoftwareVocabMt)
• #$TheoryMicrotheory -- each instance contains general assertions in a knowledge domain (e.g., #$TransportationMt ,#$ComputerSoftwareMt).
• #$DataMicrotheory -- each instance contains assertions about specific individuals (e.g., #$TransportationDataMt, #$ComputerSoftwareDataMt)
Some types of microtheories#$Microtheory
#$TheoryMicrotheory
#$DataMicrotheory
#$CounterfactualContext
#$PropositionalInformationThing
genls
genl
s
genls
genls
#$VocabularyMicrotheory genl
s
Copyright © 2002 Cycorp
Some types of microtheories#$Microtheory
#$TheoryMicrotheory
#$DataMicrotheory
#$CounterfactualContext
#$PropositionalInformationThing
genls
genl
s
genls
genls
#$VocabularyMicrotheory genl
s
•#$PropositionalInformationThing --each instance of this collection contains assertions representing the propositional content of some #$InformationBearingThing (such as a picture, text, or database table).
•#$CounterfactualContext -- each instance of this collection contains at least one assertion which is not generally taken to be true in the real world (e.g., #$TheSimpsonsMt, #$SQ77bMt)
Copyright © 2002 Cycorp
Explicitly relates a microtheory to a formula that is true in that microtheory.
• (#$ist MT FORMLA) means that the Cyc formula FORMLA is true in the microtheory MT.
Microtheory predicates: #$ist
(#$ist #$CyclistsMt (#$isa #$Lenat #$Person))
(#$ist #$NaiveStateChangeMt (#$implies (#$and (#$isa ?FREEZE #$Freezing) (#$outputsCreated ?FREEZE ?OBJ)) (#$stateOfMatter ?OBJ #$SolidStateOfMatter)))
Copyright © 2002 Cycorp
Relates two microtheories such that one of them inherits the assertions in the other; i.e., the first microtheory has access to the assertions in the second microtheory.
• (#$genlMt MT-1 MT-2) means that every assertion which is true in MT-2 is also true in MT-1.
• #$genlMt is transitive.
Microtheory predicates: #$genlMt
•(#$genlMt #$TransportationMt #$NaivePhysicsMt)
•(#$genlMt #$ModernMilitaryTacticsMt #$ModernMilitaryVehiclesMt)
•(#$genlMt #$EconomyMt #$TransportationMt)
Copyright © 2002 Cycorp
#$genlMt
Microtheory predicates, cont’d.
#$BaseKB
#$NaiveSpatialMt
#$NaivePhysicsMt#$NaturalGeographyMt
genlMtgen
lMt
genl
Mt
#$MovementMt
genlM
t
genlMt
#$TransportationMt
genl
Mt
Copyright © 2002 Cycorp
#$PredicateA sentence built from a predicate is either True or False.
• (#$mother #$ChelseaClinton #$HillaryClinton)
is True
• (#$physicalParts #$TheWhiteHouse
#$TheLincolnMemorial)
is False
Predicates are thus truth-functional relations.
Copyright © 2002 Cycorp
Argument TypesExample 1
• (#$arg1Isa #$mother #$Animal) • (#$arg2Isa #$mother #$FemaleAnimal)
Thus ,
(#$mother #$ChelseaClinton #$HillaryClinton)
is a well-formed sentence because :
(#$isa #$ChelseaClinton #$Animal) and
(#$isa #$HillaryClinton #$FemaleAnimal)
Copyright © 2002 Cycorp
Second-Order Predicates
• Sometimes we want to make statements about predicates themselves.
• This requires “second-order” predicates, which can take predicates as arguments.
Examples : #$arg1Isa, #$arity, #$isa• Thus in (#$arity #$mother 2), #$arity takes the predicate #$mother as its first argument.
• Some second-order predicates are used to relate CycL predicates to one another within a predicate hierarchy . . .
Copyright © 2002 Cycorp
The Form and Content Of The Knowledge Base
• The main advantage of Cyc over other systems for representing knowledge is its use of a formal language in which inferential connections between concepts and statements are encoded in a machine accessible way.
• The content of the Knowledge Base comprises:– A vast taxonomy of concepts and relations– A rich formal representation of their interconnections
Copyright © 2002 Cycorp
Arrangement, by Generality
Facts(Database)
Facts(Database)
UpperOntology
Core Theories
Domain-Specific Theories
Upper Ontology: Abstract Concepts
Core Theories: Space, Time, Causality, …
Domain-Specific Theories
Facts: Instances
KnowledgeBaseLayers
Copyright © 2002 Cycorp
Arrangement, by Generality
Facts(Database)
Facts(Database)
UpperOntology
Core Theories
Domain-Specific Theories
EVENT TEMPORAL-THING INDIVIDUAL THING
Upper Ontology: Abstract Concepts
KnowledgeBaseLayers
Copyright © 2002 Cycorp
Arrangement, by Generality
Facts(Database)
Facts(Database)
UpperOntology
Core Theories
Domain-Specific Theories
EVENT TEMPORAL-THING INDIVIDUAL THING
For all events a and b, a causes b implies a precedes b
Upper Ontology: Abstract Concepts
Core Theories: Space, Time, Causality, …KnowledgeBaseLayers
Copyright © 2002 Cycorp
Arrangement, by Generality
Facts(Database)
Facts(Database)
UpperOntology
Core Theories
Domain-Specific Theories
EVENT TEMPORAL-THING INDIVIDUAL THING
For all events a and b, a causes b implies a precedes b
For any mammal m and any anthrax bacteria a, m’s being exposed to a causes m to be infected by a.
Upper Ontology: Abstract Concepts
Core Theories: Space, Time, Causality, …
Domain-Specific Theories
KnowledgeBaseLayers
Copyright © 2002 Cycorp
Facts(Database)
Facts(Database)
UpperOntology
Core Theories
Domain-Specific Theories
EVENT TEMPORAL-THING INDIVIDUAL THING
For all events a and b, a causes b implies a precedes b
For any mammal m and any anthrax bacteria a, m’s being exposed to a causes m to be infected by a.
John is a person infected by anthrax.
Upper Ontology: Abstract Concepts
Core Theories: Space, Time, Causality, …
Domain-Specific Theories
Facts: Instances
KnowledgeBaseLayers
Arrangement, by Generality
Copyright © 2002 Cycorp
#$Thing
#$Intangible#$Individual
#$SetOrCollection#$TemporalThing
#$SpatialThing-Localized
#$ExistingStuffType
#$ExistingObjectType
#$Event
#$PartiallyTangible#$Collection
#$genls#$typeGenls#$disjointWith
Some Top Level Collections
Copyright © 2002 Cycorp
#$Dog (the collection of all dogs)
#$isa: #$OrganismClassificationType #$BiologicalTaxon #$BiologicalSpecies #$DomesticatedAnimalType
#$genls: #$CanineAnimal
Copyright © 2002 Cycorp
45 Collections of which #$Dog” is a Specialization
Agent Agent-Generic AirBreathingVertebrate Animal AnimalBLO BilateralObject BiologicalLivingObject CanineAnimal Carnivore CarnivoreOrder ChordataPhylum Coelomates Container-Underspecified Dog EukaryoticOrganism Eutheria FrontAndBackSidedObject Heterotroph HexalateralObject Homeotherm HumanScaleObject Individual IndividualAgent LeftAndRightSidedObject Location-Underspecified Mammal NaturalTangibleStuff NonPersonAnimal OrganicStuff Organism-Whole PartiallyTangible PerceptualAgent Region-Underspecified SentientAnimal SolidTangibleThing SomethingExisting SpatialThing SpatialThing-Localized System-Generic TemporalThing TerrestrialOrganism Thing TopAndBottomSidedObject Trajector-Underspecified Vertebrate
Copyright © 2002 Cycorp
11 Collections of which#$Dog is an Instance
#$OrganismClassificationType #$ConventionalClassificationType #$ExistingObjectType #$TemporalStuffType #$ObjectType #$Collection #$SetOrCollection#$MathematicalThing#$MathematicalOrComputationalThing#$Intangible#$PartiallyIntangible#$Thing
Copyright © 2002 Cycorp
Relations Between Temporal Things
– #$temporalBoundsIntersect
– #$temporallyIntersects
– #$startsAfterStartingOf
– #$endsAfterStartingOf
– #$endsAfterEndingOf
– #$startingDate
– #$temporallyContains
– #$temporallyCooriginating
– #$temporalBoundsContain
– #$temporalBoundsIdentical
– #$startsDuring
– #$overlapsStart
– #$startingPoint
– #$simultaneousWith
– #$after
Copyright © 2002 Cycorp
Spatial Properties and Relations
• Surfaces, Portals and Cavities
• Shape Attributes (63)• Types of Spatial
Symmetry • Direction and
Orientation Vocabulary
• Relative Positions of Objects
• Nearness and Location• Being Between• ‘In-’ Predicates (~ 60)• Connections Predicates
(~ 65)• Mereological Relations
Copyright © 2002 Cycorp
Senses of ‘In’• Does part of the inner
object stick out of the container?– None of it. -- Try
#$in-ContCompletely
– Yes -- Try
#$in-ContPartially
• If the container were turned around could the contained object fall out? – Yes -- Try
#$in-ContOpen
– No -- Try
#$in-ContClosed
Copyright © 2002 Cycorp
Senses of ‘In’ Is it attached to the
inside of the outer object?– Yes -- Try
#$connectedToInside
Can it be removed, if enough force is used,
without damaging either object?
– Yes -- Try #$in-Snugly or #$screwedIn
Does the inner object stick into the outer
object?
– Yes -- Try #$sticksInto
Copyright © 2002 Cycorp
Senses of ‘Part’
– #$parts– #$intangibleParts– #$subInformation– #$subEvents– #$physicalDecompositions– #$physicalPortions
– #$physicalParts– #$externalParts– #$internalParts– #$anatomicalParts– #$constituents– #$ingredients
Copyright © 2002 Cycorp
Some Events Types
• #$PhysicalStateChangeEvent
• #$TemperatureChangingProcess
• #$BiologicalDevelopmentEvent
• #$ChangingDeviceState
• #$CuttingNails
• #$Cracking
• #$Carving
• #$ShapeChangeEvent
• #$MovementEvent
• #$GivingSomething
• #$DiscoveryEvent
• #$Buying
• #$Thinking
• #$Baking
• #$Singing
• #$PumpingFluid
Copyright © 2002 Cycorp
Roles and ActorSlots(the world’s largest collection)
• Agency or initiating an event
• Objects acted on or changed
• Objects created or destroyed
• Facilitating objects or stuff in an event
• Slots of motion and location
• Instruments• Beneficiary/maleficiary• Specialized actor roles,
like #$plaintiffs
Copyright © 2002 Cycorp
Roles and ActorSlots“Moe clobbered Curly with the British scepter.”
• Relations between an event and its participants
• Cyc has over 200 role and ActorSlot predicates
Clobbering14
#$performedBy
Curly TheUKScepter
Moe
#$victim #$deviceUsed
Copyright © 2002 Cycorp
• This represents a particular clobbering event, not clobberings in general.
Clobbering14
performedBy
Curly TheUKScepter
Moe
victims instruments
Roles and ActorSlots“Moe clobbered Curly with the British scepter.”
Copyright © 2002 Cycorp
Roles in events and subevents
• A product of one subevent of the Krebs Process is the input to another.
• Hence, different ActorSlot predicates.
SubProcessBSubProcessA
#$inputsDestroyed#$outputsCreated
Krebs Process
Copyright © 2002 Cycorp
Information
• Information-Bearing Things– Books, web-page copies, radio broadcasts,
utterances
• Abstract strings, characters• Propositional Content• Conceptual Works
Copyright © 2002 Cycorp
“‘ T i s M o b y D i c k !”
(#$thereExists ?SEE (#$and (#$isa ?SEE Seeing) (#$objectPerceived ?SEE #$MobyDick) (#$perceiver ?SEE #$CaptainAhab)))
AbstractInformationStructure(AIS)
PropositionalInformationThing(PIT)
InformationBearingThing(IBT)
Copyright © 2002 Cycorp
“‘ T i s M o b y D i c k !”
(and (isa ?SEE Seeing) (objectPercieved ?SEE MobyDick) (perceiver ?SEE CaptainAhab))
PropositionalInformationThing(PIT)
AbstractInformationStructure(AIS)
InformationBearingThing(IBT)
Copyright © 2002 Cycorp
“‘ T i s M o b y D i c k !”
(and (isa ?SEE Seeing) (objectPercieved ?SEE MobyDick) (perceiver ?SEE CaptainAhab))
#$AbstractInformationStructure(AIS)
#$Proposition and #$PropositionalInformationThing
(PIT)
#$InformationBearingThing(IBT)
#$ConceptualWork(CW)
Copyright © 2002 Cycorp
PropositionalInformationThing(PIT)
InformationBearingThing(IBT)
ConceptualWork(CW)
AbstractInformationStructure(AIS)
Copyright © 2002 Cycorp
PropositionalInformationThing(PIT)
InformationBearingThing(IBT)
ConceptualWork(CW)
instantiationOfCW
AbstractInformationStructure(AIS)
Copyright © 2002 Cycorp
AbstractInformationStructure(AIS)
PropositionalInformationThing(PIT)
InformationBearingThing(IBT)
ConceptualWork(CW)
textOfIBT instantiationOfCW
InfoStructureOfCW
#$infoStructureRepresents
ContainsInfo-Propositional-CW
PITOfIBTFn
Copyright © 2002 Cycorp
Emotion
• Feeling Attributes Types– #$Abhorrence– #$Adulation– #$Relaxed-Feeling– #$Gratitude– #$Anticipation-Feeling– Over 120 of these
• Relations Pertaining to emotions
– #$contraryFeelings– #$feelsTowardsObject– #$appropriateEmotion – #$feelsTowardsPersonType– #$actionExpressesFeeling
Copyright © 2002 Cycorp
Propositional Attitudes Relations Between Agents and Propositions
• #$goals• #$intends• #$desires• #$hopes• #$expects• #$beliefs
• #$opinions • #$knows• #$rememberedProp• #$perceivesThat• #$seesThat• #$tastesThat
Copyright © 2002 Cycorp
Biology
• Organisms classified by:– Taxon
– Habitat
– Source of Nutrients
• Some scientific, #$ChordataPhylum, some not, #$Worm
• Organism Anatomy– Gross Anatomy
– Cell biology
– Physiological Processes
• Life stages
Copyright © 2002 Cycorp
Materials
• Common Substances• Attributes of Materials• States Of Matter• Solutions
• Electrical Conductivity • Thermal Conductivity• Structural Attributes• Tangible Attributes
Copyright © 2002 Cycorp
Devices
• Specializations Of #$PhysicalDevice
• Device States• Device Actions• Device Predicates• Device Purposes
Copyright © 2002 Cycorp
Weather
• Weather Attributes– #$ClearWeather– (#$LowAmountFn
#$Raininess)• Weather Events– #$TornadoAsEvent– #$SnowProcess
• Weather Objects– #$CloudInSky– #$TornadoAsObject
Copyright © 2002 Cycorp
Geography
• Geopolitical Entities• Addresses• Specific Ethnic and Language information• Borders• Districts, States, etc. • Seas, islands, straits,etc.
Copyright © 2002 Cycorp
Inference uses Deduction:Rules + Facts
Deduction - rule fact(s) new fact(#$loves #$Hamlet #$Gertrude)
“Rules” - general, variables(#$implies (#$mother ?PERSON ?MOTHER) (#$loves ?PERSON ?MOTHER))
“Facts” - specific, no variables(#$mother #$Hamlet #$Gertrude)
Copyright © 2002 Cycorp
The Resolution Principle
(#$and (#$knows #$Hamlet ?WHO) (#$loves #$Hamlet ?WHO))
(#$implies (#$mother ?PERSON ?MOTHER) (#$loves ?PERSON ?MOTHER))
Query Rule
Resolution Principle : “Unify, Substitute, Merge”
Copyright © 2002 Cycorp
The Resolution Principle: Unify
(#$and (#$knows #$Hamlet ?WHO) (#$loves #$Hamlet ?WHO))
(#$implies (#$mother ?PERSON ?MOTHER) (#$loves ?PERSON ?MOTHER))
Resolution Principle : “Unify, Substitute, Merge”
Query Rule
Pivot Literals
Copyright © 2002 Cycorp
(#$and (#$knows #$Hamlet ?WHO) (#$loves #$Hamlet ?WHO))
(#$implies (#$mother ?PERSON ?MOTHER) (#$loves ?PERSON ?MOTHER))
Most General Unifier#$Hamlet / ?PERSON?WHO / ?MOTHER
Resolution Principle : “Unify, Substitute, Merge”
Query Rule
The Resolution Principle: Unify
(#$loves #$Hamlet ?WHO))=(#$loves #$Hamlet ?WHO))
Copyright © 2002 Cycorp
The Resolution Principle: Substitute
(#$and (#$knows #$Hamlet ?WHO) (#$loves #$Hamlet ?WHO))
(#$implies (#$mother ?PERSON ?MOTHER) (#$loves ?PERSON ?MOTHER))
Most General Unifier#$Hamlet / ?PERSON?WHO / ?MOTHER
Resolution Principle : “Unify, Substitute, Merge”Query Rule
(#$and (#$knows #$Hamlet ?WHO) (#$loves #$Hamlet ?WHO))
(#$implies (#$mother #$Hamlet ?WHO) (#$loves #$Hamlet ?WHO))
Substituted Query Substituted Rule
Copyright © 2002 Cycorp
The Search Tree
Query/root
Possible Answers:Branches/Child nodes
“Who does Hamlet know and love?”
Copyright © 2002 Cycorp
Justifying the Answers
Query
Goal!
Logical Deduction
with the Most General Unifier
Copyright © 2002 Cycorp
The Halting Problem
millions of facts+ tens of thousands of rulesinfinite possibilities
Copyright © 2002 Cycorp
Breadth-first Search: advantage and disadvantage
0-step
1-step
2-step
n-step
+ Deal with simplest proofs first- Infinite fan-out is common
Copyright © 2002 Cycorp
Cyc is Life in the Big City
Searching in Cyc:~100,000 constants
~1 million assertions
Copyright © 2002 Cycorp
Inference is Resource-bounded
Resource Bounds :– quit after NUMBER of answers– quit after TIME seconds– ignore any proof using more than
BACKCHAIN rules– ignore any proof using more than DEPTH steps
Copyright © 2002 Cycorp
Inference Uses Mts for Consistency
WorldMythologyMt
•(genls Vampire IntelligentAgent)
•(isa LochNessMonster Reptile)
MainstreamAmericanCultureMt
•(genls Vampire MythologicalThing)
•(isa LochNessMonster MythologicalThing)
In the Mainstream AmericanCultureMt, •Vampire is a kind of mythological thing.•The Loch Ness Monster is a mythological thing.
In the WorldMythologyMt,•Vampire is a kind of intelligent agent.•The Loch Ness Monster is a reptile.
Copyright © 2002 Cycorp
Mts Inherit from More General Mts Using #$genlMt
UniversalVocabularyMt
MainstreamAmericanCultureMt
UnitedStatesSocialLifeMt
genlMt
HumanActivitiesMt
genlMt
genlMt
genlMt
WorldMythologyMt