building a system that can learn by reading

55
1 Building a System that can Learn by Reading Kevin Livingston PhD Candidate Cognitive Systems Division EECS Department Northwestern University Presented at University of Dayton November 3, 2006 As part of the Computer Science Research Colloquium Series

Upload: amber

Post on 14-Jan-2016

38 views

Category:

Documents


0 download

DESCRIPTION

Building a System that can Learn by Reading. Kevin Livingston PhD Candidate Cognitive Systems Division EECS Department Northwestern University Presented at University of Dayton November 3, 2006 As part of the Computer Science Research Colloquium Series. Text. Understanding Systems. Text - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Building a System that can Learn by Reading

1

Building a System that canLearn by Reading

Kevin LivingstonPhD Candidate

Cognitive Systems DivisionEECS Department

Northwestern University

Presented at University of Dayton November 3, 2006As part of the Computer Science Research Colloquium Series

Page 2: Building a System that can Learn by Reading

2

Understanding Systems

Knowledge Base

Text

Text Understanding

System QuestionAnswering

Explanation

Reasoning

Page 3: Building a System that can Learn by Reading

3

Examples of Intelligent Systems

• Digital Assistants

• Conversational Agents– Imagine a video game character that actually

talked to you

• Intelligent Games

• Search

• Question Answering

Page 4: Building a System that can Learn by Reading

4

Modeling and Using Knowledge

• Frames (Minsky)

• Scripts (Shank)

• Qualitative Reasoning (Forbus)

Page 5: Building a System that can Learn by Reading

5

Semantic Memory

• Ontology– Doctors are people– Bombings are events

• Semantics– Doctors have patients– Bombings have targets

Page 6: Building a System that can Learn by Reading

6

Episodic Memory

• Memory Instances– Madrid is a city– Madrid is in Spain– The Madrid Bombings is a terrorist attack– The Madrid Bombings occurred on March 11,

2004– Al Qaida is a terrorist organization

Page 7: Building a System that can Learn by Reading

7

Modeling Reasoning

• Logic Based Systems– Cyc (Cycorp; Austin, TX)– Fire (Qualitative Reasoning Group (QRG);

Northwestern)

• Statistical• Bayesian• Markovian• Neural Nets

Page 8: Building a System that can Learn by Reading

8

Knowledge Bases

Knowledge Base

Text

Text Understanding

System QuestionAnswering

Explanation

Reasoning

Page 9: Building a System that can Learn by Reading

9

Available Knowledge Bases• Information Retrieval (IR) techniques

– mine information from Internet (MUC and TREC)

• Open Mind Common Sense (OMCS)– Sentences collected from Internet contributors– Mined for knowledge

• Knowledge Machine (KM)– Frame based

• ResearchCyc– Latest release ~3,000,000 assertions– Predicate Logic in CycL

• First order and some second order constructs

Page 10: Building a System that can Learn by Reading

10

ResearchCyc

• Bombings are Attacks and Attacks are Events(genls Bombing AttackOnObject)

(genls AttackOnObject Event)

• The Madrid terrorist attack was a bombing(isa

TerroristAttack-September-8-2003-Madrid

Bombing)

• Al Qaida is an Islamist Terrorist Group(isa AlQaida TerroristGroup-Islamist)

Page 11: Building a System that can Learn by Reading

11

ResearchCyc (cont.)

• Location of an event

(eventOccursAt

TerroristAttack-September-8-2003-Madrid

CityOfMadridSpain) • Perpetrator of an event(perpetrator

TerroristAttack-September-8-2003-Madrid

AlQaida)

Page 12: Building a System that can Learn by Reading

12

ResearchCyc (cont.)

• Person being killed(organismKilled SpaceShuttleChallengerDisaster ChristaMcAuliffe)

• Deaths caused by an attack(deathToll

TerroristAttack-September-8-2003-Madrid

Person190)

Page 13: Building a System that can Learn by Reading

13

Building Knowledge Bases

• Slow and Tedious– years to grow Cyc from 1.2M to 3M assertions

• Requires Training– measured in weeks+, for GUI tools (SHAKEN)

• Expensive– Project Halo estimates $10,000 per page!

of AP level Chemistry content

Page 14: Building a System that can Learn by Reading

14

Available Information

• Encyclopedias

• Newspapers

• Online sources

• Print

Page 15: Building a System that can Learn by Reading

15

A Better Way?Teach the Computer to Read

Page 16: Building a System that can Learn by Reading

16

What we want to Read

• Episodic Knowledge– New people– New events

• General Knowledge– “the heart is a pump”

Page 17: Building a System that can Learn by Reading

17

Standard Model forNatural Language Processing

Dictionary

Text

POSTagging

Grammar

SyntacticParser

SemanticInterpreter

Semantics

Tagged Text

Page 18: Building a System that can Learn by Reading

18

“Time flies like an arrow.”

Page 19: Building a System that can Learn by Reading

19

“Time flies like an arrow.”• Time moves quickly just like an arrow does

Page 20: Building a System that can Learn by Reading

20

“Time flies like an arrow.”• Time moves quickly just like an arrow does

• (You should) time flies like you would an arrow

Page 21: Building a System that can Learn by Reading

21

“Time flies like an arrow.”• Time moves quickly just like an arrow does

• (You should) time flies like you would an arrow

• Time flies in the same way that an arrow would (time them)

Page 22: Building a System that can Learn by Reading

22

“Time flies like an arrow.”• Time moves quickly just like an arrow does

• (You should) time flies like you would an arrow

• Time flies in the same way that an arrow would (time them)

• Time those flies that are like arrows;

Page 23: Building a System that can Learn by Reading

23

“Time flies like an arrow.”• Time moves quickly just like an arrow does

• (You should) time flies like you would an arrow

• Time flies in the same way that an arrow would (time them)

• Time those flies that are like arrows;

• A type of flying insect, "time-flies," enjoy arrows (compare Fruit flies like a banana.)

Page 24: Building a System that can Learn by Reading

24

Picture ofGeorge Burns

AndGrace Allen

Page 25: Building a System that can Learn by Reading

25

Grace,Those are beautiful flowers.

Picture ofGeorge Burns

AndGrace Allen

Page 26: Building a System that can Learn by Reading

26

Grace,Those are beautiful flowers.

Where did they come

from?

Picture ofGeorge Burns

AndGrace Allen

Page 27: Building a System that can Learn by Reading

27

Don’t you remember, George?

Picture ofGeorge Burns

AndGrace Allen

Page 28: Building a System that can Learn by Reading

28

Picture ofGeorge Burns

AndGrace Allen

You said that if I went to visit Clara Bagley

in the hospital I should be sure to take her flowers.

So when she wasn’t looking,

I did.

Page 29: Building a System that can Learn by Reading

29

Picture ofGeorge Burns

AndGrace Allen

Page 30: Building a System that can Learn by Reading

30

Common Sense

• “You said that if I went to visit Clara Bagley in the hospital I should be sure to take her flowers. So, when she wasn't looking, I did.”

– take flowers from her– take flowers to her

Page 31: Building a System that can Learn by Reading

31

Picture ofElevatorOperator

Page 32: Building a System that can Learn by Reading

32

Down?

Picture ofElevatorOperator

Page 33: Building a System that can Learn by Reading

33

ThatWay!

Picture ofElevatorOperator

Page 34: Building a System that can Learn by Reading

34

Context in the Environment

“Down?”

What does the question mean?– Which way is down?– Are you going down?

Page 35: Building a System that can Learn by Reading

35

Language UnderstandingGoals include understanding:

– The meaning of the text– How it fits into what is known– The purpose of being told

Requires:– Common sense knowledge– Awareness of context

Our model is to get to knowledge as quickly as possible, with as few intermediate steps as possible.

Page 36: Building a System that can Learn by Reading

36

Page 37: Building a System that can Learn by Reading

37

Reader ExampleAn attack occurred in MadridThe bombing killed 190 peopleThe bombing was perpetrated by Al-Qaida

(eventOccursAt TerroristAttack-September-8-2003-Madrid CityOfMadridSpain)(perpetrator TerroristAttack-September-8-2003-Madrid AlQaida)(deathToll TerroristAttack-September-8-2003-Madrid Person 190)(isa TerroristAttack-September-8-2003-Madrid Bombing)

Page 38: Building a System that can Learn by Reading

38

Lexical Processing

“An attack occurred in Madrid.”• “attack”(singular Attack-TheWord “attack”)(denotation Attack-TheWord CountNoun 0

AttackOnObject)(isa ?x AttackOnObject)

• “Madrid”(placeName-Standard CityOfMadridSpain

"Madrid")(isa CityOfMadridSpain City)

Page 39: Building a System that can Learn by Reading

39

Rule Pattern

• Pattern:(isa ?event Event)

Occur-TheWord

In-TheWord

(isa ?location GeographicLocation)

• Results:(eventOccursAt ?event ?

location)

Page 40: Building a System that can Learn by Reading

40

Pattern Matching• Pattern:

(isa ?event Event) Occur-TheWordIn-TheWord (isa ?location GeographicLocation)

• Input:– “An”– “attack” (isa ?x AttackOnObject)

– “occurred” Occur-TheWord

– “in” In-TheWord– “Madrid” (isa CityOfMadridSpain City)

Page 41: Building a System that can Learn by Reading

41

Rule Completion“An attack occurred in Madrid.”• Pattern:

(isa ?event Event) Occur-TheWordIn-TheWord (isa ?location GeographicLocation)

• Results:(eventOccursAt ?event ?location)

• Constraints:(isa ?event AttackOnObject)

• Bindings from Reading:((?location . CityOfMadridSpain))

Page 42: Building a System that can Learn by Reading

42

Remindings“An attack occurred in Madrid.”• Pattern: (isa ?event Event)

Occur-TheWordIn-TheWord (isa ?location GeographicLocation)

• Results:(eventOccursAt ?event ?location)

• Constraints:(isa ?event AttackOnObject)

• Bindings from Reading:((?location . CityOfMadridSpain))

• Remindings from Memory:((?event . TerrorAttack-Sept8-2003-Madrid))

Page 43: Building a System that can Learn by Reading

43

Coreference ResolutionAn attack occurred in Madrid. The bombing was

perpetrated by Al-Qaida.

Page 44: Building a System that can Learn by Reading

44

Coreference ResolutionAn attack occurred in Madrid

Results: (eventOccursAt?event?location)

Constraints:(isa ?event AttackOnObject)

Bindings from Reading:((?location .

CityOfMadridSpain))Remindings from Memory:((?event . TerrorAttack-

Sept8-2003-Madrid))

The bombing was perpetrated by Al-Qaida

Results: (perpetrator?action

?agent)

Constraints:(isa ?action Bombing)

Bindings from Reading:((?agent . AlQaida))

Page 45: Building a System that can Learn by Reading

45

Coreference ResolutionAn attack occurred in Madrid

Results: (eventOccursAt?event?location)

Constraints:(isa ?event AttackOnObject)

Bindings from Reading:((?location .

CityOfMadridSpain))Remindings from Memory:((?event . TerrorAttack-

Sept8-2003-Madrid))

The bombing was perpetrated by Al-Qaida

Results: (perpetrator?action

?agent)

Constraints:(isa ?action Bombing)

Bindings from Reading:((?agent . AlQaida))

Page 46: Building a System that can Learn by Reading

46

Coreference ResolutionAn attack occurred in Madrid

Results: (eventOccursAt?event?location)

Constraints:(isa ?event AttackOnObject)

Bindings from Reading:((?location .

CityOfMadridSpain))Remindings from Memory:((?event . TerrorAttack-

Sept8-2003-Madrid))

The bombing was perpetrated by Al-Qaida

Results: (perpetrator?action

?agent)

Constraints:(isa ?action Bombing)

Bindings from Reading:((?agent . AlQaida))

Page 47: Building a System that can Learn by Reading

47

Coreference ResolutionAn attack occurred in Madrid

Results: (eventOccursAt?event?location)

Constraints:(isa ?event AttackOnObject)

Bindings from Reading:((?location .

CityOfMadridSpain))Remindings from Memory:((?event . TerrorAttack-

Sept8-2003-Madrid))

The bombing was perpetrated by Al-Qaida

Results: (perpetrator?action

?agent)

Constraints:(isa ?action Bombing)

Bindings from Reading:((?agent . AlQaida))

Page 48: Building a System that can Learn by Reading

48

Coreference ResolutionAn attack occurred in Madrid

Results: (eventOccursAt?event?location)

Constraints:(isa ?event AttackOnObject)

Bindings from Reading:((?location .

CityOfMadridSpain))Remindings from Memory:((?event . TerrorAttack-

Sept8-2003-Madrid))

The bombing was perpetrated by Al-Qaida

Results: (perpetrator?action

?agent)

Constraints:(isa ?action Bombing)

Bindings from Reading:((?agent . AlQaida))

Page 49: Building a System that can Learn by Reading

49

Coreference

• Refer to a more general or specific type– “bombing” and “attack”

• Consistent with being the same,

have a known shared instance in memory– “doctor” and “father of four”

Page 50: Building a System that can Learn by Reading

50

Tracking Ambiguity

• Words are ambiguous– “Bush”

• A shrubbery?• A president? Which one?

• Sentences and Phrases– “take her flowers”– “Iraq borders Iran on the North.”

• Intension– “Down?”– “Where is Baghdad?”

Page 51: Building a System that can Learn by Reading

51

% of sentences processed under time

0

10

20

30

40

50

60

70

80

90

100

1 10 100 1000 10000 100000 1E+06 1E+07

time (ms)

% o

f co

rpu

s se

nte

nce

s p

roce

ssed

RDE

SLE

WLE

Page 52: Building a System that can Learn by Reading

52

What Learning Reader can Read

• Information about existing instances– “the 2003 Madrid bombing”

• Information extending existing instances– “the attack killed 190 people”

• Information representing new instances– “An attack occurred in Al Anbar.”

Page 53: Building a System that can Learn by Reading

53

What Learning Reader can’t Read

• Larger patterns– Scripts: “A bomb went off.”

“6 people were arrested.”

• Generalizations“All countries have political leaders.”

“The heart is a pump.”

Page 54: Building a System that can Learn by Reading

54

Dr. Christopher K. RiesbeckDr. Ken ForbusDr. Larry BirnbaumAbhishek Sharma

Dr. Jennifer SeitzerDr. Saverio Perugini

grant HR0011-04-1-0051

Acknowledgements

Page 55: Building a System that can Learn by Reading

55

For more Information

http://cs.northwestern.edu/~livingston/http://cs.northwestern.edu/~livingston/talks.html

Kevin LivingstonPhD Candidate

Cognitive Systems DivisionEECS Department

Northwestern University