47
Never Ending Language Learning Mausam Slides by Tom Mitchell, Partha Pratim Talukdar

Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Never Ending Language Learning

Mausam

Slides by Tom Mitchell, Partha Pratim Talukdar

Page 2: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Tenet 1:

Understanding requires a belief system

We’ll never produce natural language

understanding systems until we have

systems that react to arbitrary sentences

by saying one of:

• I understand, and already knew that

• I understand, and didn’t know, but accept it

• I understand, and disagree because …

Page 3: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Tenet 2:

We’ll never really understand learning

until we build machines that

• learn many different things,

• over years,

• and become better learners over time.

Page 4: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

NELL: Never-Ending Language Learner

Inputs:

• initial ontology (categories and relations)

• dozen examples of each ontology predicate

• the web

• occasional interaction with human trainers

The task:

• run 24x7, forever

• each day:

1. extract more facts from the web to populate the ontology

2. learn to read (perform #1) better than yesterday

Page 5: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

NELL today

Running 24x7, since January, 12, 2010

Result:

• KB with > 90 million candidate beliefs, growing daily

• learning to read better each day

• learning to reason, as well as read

• automatically extending its ontology

Page 6: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Globe and Mail

Stanley

Cup

hockey

NHL

Toronto

CFRB

Wilson

playhired

wonMaple Leafs

home town

city

paper

league

Sundin

Milson

writer

radio

Air Canada

Centre

team stadium

Canada

city

stadium

politician

country

Miller

airport

member

Toskala

Pearson

Skydome

Connaught

Sunnybrook

hospital

city company

skates helmet

uses equipment

wonRed

Wings

Detroit

hometown

GM

city company

competeswith

Toyota

plays in

league

Prius

Corrola

createdHino

acquired

automobile

economicsector

city

stadium

NELL knowledge fragmentclimbing

football

uses equipment

Page 7: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

How does NELL work?

Page 8: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Continually Learning Extractors

Basic NELL Architecture

Knowledge Base

(latent variables)

Text

Context

patterns

(CPL)

HTML-URL

context

patterns

(SEAL)

Morphology

classifier

(CML)

Beliefs

Candidate

Beliefs

Evidence

Integrator

Page 9: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Semi-Supervised Bootstrap Learning

Paris

Pittsburgh

Seattle

Montpelier

mayor of arg1

live in arg1

San Francisco

Berlin

denial

arg1 is home of

traits such as arg1

it’s underconstrained!!

anxiety

selfishness

London

Learn which

noun phrases

are cities:

Page 10: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

hard

(underconstrained)

semi-supervised

learning problem

Key Idea 1: Coupled semi-supervised training

of many functions

much easier (more constrained)

semi-supervised learning problem

person

noun phrase

Page 11: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

NP:

person

Type 1 Coupling: Co-Training, Multi-View Learning

Supervised training of 1 function:

Minimize:

Page 12: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

NP:

person

Type 1 Coupling: Co-Training, Multi-View Learning

Coupled training of 2 functions:

Minimize:

Page 13: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

NP:

person

Type 1 Coupling: Co-Training, Multi-View Learning[Blum & Mitchell; 98]

[Dasgupta et al; 01 ]

[Ganchev et al., 08]

[Sridharan & Kakade, 08]

[Wang & Zhou, ICML10]

Page 14: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

team

person

athletecoach

sport

NP

athlete(NP) person(NP)

athlete(NP) NOT sport(NP)

NOT athlete(NP) sport(NP)

Type 2 Coupling: Multi-task, Structured Outputs[Daume, 2008]

[Bakhir et al., eds. 2007]

[Roth et al., 2008]

[Taskar et al., 2009]

[Carlson et al., 2009]

Page 15: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

team

person

NP:

athletecoach

sport

NP text

context

distribution

NP

morphology

NP HTML

contexts

Multi-view, Multi-Task Coupling

Page 16: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

coachesTeam(c,t)playsForTeam(a,t) teamPlaysSport(t,s)

playsSport(a,s)

NP1 NP2

Type 3 Coupling: Learning Relations

Page 17: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

team

coachesTeam(c,t)playsForTeam(a,t) teamPlaysSport(t,s)

playsSport(a,s)

person

NP1

athlete

coach

sport

team

person

NP2

athlete

coach

sport

playsSport(NP1,NP2) athlete(NP1), sport(NP2)

Type 3 Coupling: Argument Types

over 2500 coupled

functions in NELL

Page 18: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

NELL: Learned reading strategies

Plays_Sport(arg1,arg2):

arg1_was_playing_arg2 arg2_megastar_arg1 arg2_icons_arg1

arg2_player_named_arg1 arg2_prodigy_arg1

arg1_is_the_tiger_woods_of_arg2 arg2_career_of_arg1

arg2_greats_as_arg1 arg1_plays_arg2 arg2_player_is_arg1

arg2_legends_arg1 arg1_announced_his_retirement_from_arg2

arg2_operations_chief_arg1 arg2_player_like_arg1

arg2_and_golfing_personalities_including_arg1 arg2_players_like_arg1

arg2_greats_like_arg1 arg2_players_are_steffi_graf_and_arg1

arg2_great_arg1 arg2_champ_arg1 arg2_greats_such_as_arg1

arg2_professionals_such_as_arg1 arg2_hit_by_arg1 arg2_greats_arg1

arg2_icon_arg1 arg2_stars_like_arg1 arg2_pros_like_arg1

arg1_retires_from_arg2 arg2_phenom_arg1 arg2_lesson_from_arg1

arg2_architects_robert_trent_jones_and_arg1 arg2_sensation_arg1

arg2_pros_arg1 arg2_stars_venus_and_arg1 arg2_hall_of_famer_arg1

arg2_superstar_arg1 arg2_legend_arg1 arg2_legends_such_as_arg1

arg2_players_is_arg1 arg2_pro_arg1 arg2_player_was_arg1

arg2_god_arg1 arg2_idol_arg1 arg1_was_born_to_play_arg2

arg2_star_arg1 arg2_hero_arg1 arg2_players_are_arg1

arg1_retired_from_professional_arg2 arg2_legends_as_arg1

arg2_autographed_by_arg1 arg2_champion_arg1 …

Page 19: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Results

• 6 Months after launch:

– For ¾ of relations and categories, 90-99% accuracy.

– For 1/4 of relations and categories, 25-60% accuracy.

• After that, weekly sessions held for human

review of labels, intervening when blatant errors

arose.

• Twitter handle for getting free human feedback

– Now, anyone can interact via NELL’s website:

– http://rtw.ml.cmu.edu/rtw/

Page 21: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

NELL Architecture

Knowledge Base

(latent variables)

Text

Context

patterns

(CPL)

Orthographic

classifier

(CMC)

Beliefs

Candidate

Beliefs

Evidence

Integrator

Human

advice

Actively

search for

web text

(OpenEval)

Infer new

beliefs from

old

(PRA)

Image

classifier

(NEIL)

Ontology

extender

(OntExt)

URL specific

HTML

patterns

(SEAL)

Page 22: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

If coupled learning is the key,

how can we get new coupling constraints?

Page 23: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Key Idea 2:

Discover New Coupling Constraints

• first order, probabilistic horn clause constraints:

– learned by data mining the knowledge base

– connect previously uncoupled relation predicates

– infer new unread beliefs

– modified version of FOIL [Quinlan]

0.93 athletePlaysSport(?x,?y) athletePlaysForTeam(?x,?z)

teamPlaysSport(?z,?y)

Page 24: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Example Learned Horn Clauses

athletePlaysSport(?x,basketball) athleteInLeague(?x,NBA)

athletePlaysSport(?x,?y) athletePlaysForTeam(?x,?z)

teamPlaysSport(?z,?y)

teamPlaysInLeague(?x,NHL) teamWonTrophy(?x,Stanley_Cup)

athleteInLeague(?x,?y) athletePlaysForTeam(?x,?z),

teamPlaysInLeague(?z,?y)

cityInState(?x,?y) cityCapitalOfState(?x,?y), cityInCountry(?y,USA)

newspaperInCity(?x,New_York) companyEconomicSector(?x,media)

generalizations(?x,blog)

0.95

0.93

0.91

0.90

0.88

0.62*

Page 25: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

team

coachesTeam(c,t)playsForTeam(a,t) teamPlaysSport(t,s)

playsSport(a,s)

person

NP1

athlete

coach

sport

team

person

NP2

athlete

coach

sport

Learned Probabilistic Horn Clause Rules

0.93 playsSport(?x,?y) playsForTeam(?x,?z), teamPlaysSport(?z,?y)

Page 26: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Inference by KB Random Walks [Lao et al, EMNLP 2011]

If: x1competes

with (x1,x2)

x2economic

sector (x2, x3)

x3

Then: economic sector (x1, x3)

Page 27: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Inference by KB Random Walks [Lao et al, EMNLP 2011]

KB:

Random walk

path type:

Trained logistic function for R, where ith

feature is probability of arriving at node y

when starting at node x, and taking a random

walk along path type i

Infer Pr(R(x,y)):

?competes

with?

economic sector

?

Page 28: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Feature = Typed Path

CityInState, CityInstate-1, CityLocatedInCountry 0.8 0.32

Pittsburgh

Feature Value

Logistic

Regresssion

Weight

CityLocatedInCountry(Pittsburgh) = ? [Lao et al, EMNLP 2011]

Page 29: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Feature = Typed Path

CityInState, CityInstate-1, CityLocatedInCountry 0.8 0.32

Pittsburgh

Pennsylvania

Feature Value

Logistic

Regresssion

Weight

CityLocatedInCountry(Pittsburgh) = ? [Lao et al, EMNLP 2011]

Page 30: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Feature = Typed Path

CityInState, CityInstate-1, CityLocatedInCountry 0.8 0.32

Pittsburgh

Pennsylvania

PhiladelphiaHarisburg

…(14)

Feature Value

Logistic

Regresssion

Weight

CityLocatedInCountry(Pittsburgh) = ? [Lao et al, EMNLP 2011]

Page 31: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Feature = Typed Path

CityInState, CityInstate-1, CityLocatedInCountry 0.8 0.32

Pittsburgh

Pennsylvania

PhiladelphiaHarisburg

…(14)

U.S.

Feature Value

Logistic

Regresssion

Weight

CityLocatedInCountry(Pittsburgh) = ? [Lao et al, EMNLP 2011]

Page 32: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Feature = Typed Path

CityInState, CityInstate-1, CityLocatedInCountry 0.8 0.32

Pittsburgh

Pennsylvania

PhiladelphiaHarisburg

…(14)

U.S.

Feature Value

Logistic

Regresssion

Weight

CityLocatedInCountry(Pittsburgh) = ?

Pr(U.S. | Pittsburgh, TypedPath)

[Lao et al, EMNLP 2011]

Page 33: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Feature = Typed Path

CityInState, CityInstate-1, CityLocatedInCountry 0.8 0.32

AtLocation-1, AtLocation, CityLocatedInCountry 0.20

Pittsburgh

Pennsylvania

PhiladelphiaHarisburg

…(14)

U.S.

Feature Value

Logistic

Regresssion

Weight

CityLocatedInCountry(Pittsburgh) = ? [Lao et al, EMNLP 2011]

Page 34: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Feature = Typed Path

CityInState, CityInstate-1, CityLocatedInCountry 0.8 0.32

AtLocation-1, AtLocation, CityLocatedInCountry 0.20

Pittsburgh

Pennsylvania

PhiladelphiaHarisburg

…(14)

U.S.

Feature Value

Logistic

Regresssion

Weight

DeltaPPG

CityLocatedInCountry(Pittsburgh) = ? [Lao et al, EMNLP 2011]

Page 35: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Feature = Typed Path

CityInState, CityInstate-1, CityLocatedInCountry 0.8 0.32

AtLocation-1, AtLocation, CityLocatedInCountry 0.20

Pittsburgh

Pennsylvania

PhiladelphiaHarisburg

…(14)

U.S.

Feature Value

Logistic

Regresssion

Weight

DeltaPPG

AtlantaDallas

Tokyo

CityLocatedInCountry(Pittsburgh) = ? [Lao et al, EMNLP 2011]

Page 36: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Feature = Typed Path

CityInState, CityInstate-1, CityLocatedInCountry 0.8 0.32

AtLocation-1, AtLocation, CityLocatedInCountry 0.6 0.20

Pittsburgh

Pennsylvania

PhiladelphiaHarisburg

…(14)

U.S.

Feature Value

Logistic

Regresssion

Weight

DeltaPPG

AtlantaDallas

Tokyo

Japan

CityLocatedInCountry(Pittsburgh) = ? [Lao et al, EMNLP 2011]

Page 37: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Feature = Typed Path

CityInState, CityInstate-1, CityLocatedInCountry 0.8 0.32

AtLocation-1, AtLocation, CityLocatedInCountry 0.6 0.20

… … …

Pittsburgh

Pennsylvania

PhiladelphiaHarisburg

…(14)

U.S.

Feature Value

Logistic

Regresssion

Weight

CityLocatedInCountry(Pittsburgh) = U.S. p=0.58

DeltaPPG

AtlantaDallas

Tokyo

Japan

CityLocatedInCountry(Pittsburgh) = ? [Lao et al, EMNLP 2011]

Page 38: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Random walk inference: learned path types

CityLocatedInCountry(city, country):

8.04 cityliesonriver, cityliesonriver-1, citylocatedincountry

5.42 hasofficeincity-1, hasofficeincity, citylocatedincountry

4.98 cityalsoknownas, cityalsoknownas, citylocatedincountry

2.85 citycapitalofcountry,citylocatedincountry-1,citylocatedincountry

2.29 agentactsinlocation-1, agentactsinlocation, citylocatedincountry

1.22 statehascapital-1, statelocatedincountry

0.66 citycapitalofcountry...

7 of the 2985 paths for inferring CityLocatedInCountry

Page 39: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Random Walk Inference: ExampleRank 17 companies by probability competesWith(MSFT,X):

NELL/PRA ranking

GoogleOracleIBMAppleSAPYahoo

FacebookRedhatLenovoFedExSAS

BoeingHondaDupontLufthansaExxonPfizer

Page 40: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Random Walk Inference: ExampleRank 17 companies by probability competesWith(MSFT,X):

NELL/PRA ranking Human Ranking (9 subjs)

AppleGoogleYahooIBM

RedhatOracle

FacebookSAPSAS

LenovoBoeingHondaFedExDupontExxon

LufthansaPfizer

GoogleOracleIBMAppleSAPYahoo

FacebookRedhatLenovoFedExSAS

BoeingHondaDupontLufthansaExxonPfizer

1. Tractable

(bounded length)

2. Anytime

3. Accuracy increases

as KB grows

4. Addresses question

of how to combine

probabilities from

different horn clauses

Page 41: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Key Idea 3:

Automatically extend ontology

Page 42: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Ontology Extension

Goal:

• Add new relations to ontology

Approach:

• For each pair of categories C1, C2,

• cluster pairs of known instances, in terms of

text contexts that connect them

[Mohamed et al., EMNLP 2011]

Page 43: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Example Discovered Relations

Category Pair Frequent Instance Pairs Text ContextsSuggested

Name

MusicInstrumentMusician

sitar, George Harrisontenor sax, Stan Getz

trombone, Tommy Dorseyvibes, Lionel Hampton

ARG1 master ARG2ARG1 virtuoso ARG2ARG1 legend ARG2ARG2 plays ARG1

Master

DiseaseDisease

pinched nerve, herniated disktennis elbow, tendonitis

blepharospasm, dystonia

ARG1 is due to ARG2ARG1 is caused by ARG2

IsDueTo

CellTypeChemical

epithelial cells, surfactantneurons, serotonin

mast cells, histomine

ARG1 that release ARG2ARG2 releasing ARG1

ThatRelease

MammalsPlant

koala bears, eucalyptussheep, grassesgoats, saplings

ARG1 eat ARG2ARG2 eating ARG1

Eat

RiverCity

Seine, Paris

Nile, Cairo

Tiber river, Rome

ARG1 in heart of ARG2

ARG1 which flows through

ARG2

InHeartOf

[Mohamed et al. EMNLP 2011]

Page 44: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

NELL: sample of self-added relations

• athleteWonAward

• animalEatsFood

• languageTaughtInCity

• clothingMadeFromPlant

• beverageServedWithFood

• fishServedWithFood

• athleteBeatAthlete

• athleteInjuredBodyPart

• arthropodFeedsOnInsect

• animalEatsVegetable

• plantRepresentsEmotion

• foodDecreasesRiskOfDisease

• clothingGoesWithClothing

• bacteriaCausesPhysCondition

• buildingMadeOfMaterial

• emotionAssociatedWithDisease

• foodCanCauseDisease

• agriculturalProductAttractsInsect

• arteryArisesFromArtery

• countryHasSportsFans

• bakedGoodServedWithBeverage

• beverageContainsProtein

• animalCanDevelopDisease

• beverageMadeFromBeverage

Page 45: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

Key Idea 4: Cumulative, Staged Learning

1. Classify noun phrases (NP’s) by category

2. Classify NP pairs by relation

3. Discover rules to predict new relation instances

4. Learn which NP’s (co)refer to which latent concepts

5. Discover new relations to extend ontology

6. Learn to infer relation instances via targeted random walks

7. Learn to assign temporal scope to beliefs

8. Learn to microread single sentences

9. Vision: co-train text and visual object recognition

10. Goal-driven reading: predict, then read to

corroborate/correct

11. Make NELL a conversational agent on Twitter

12. Add a robot body to NELL

Learning X improves ability to learn Y

Page 46: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

NELL Summary

• Learning

– Coupled multi-task, multi-view semi-supervised training

• Inference

– Data mine the KB to learn inference rules

– Scalable any-time inference via random walks

• Representation

– Ontology extension

• invent new categories and relations

• combine statistical clustering with direct reading

– Infer millions of latent concepts from observable text

• Curriculum

– learn easiest things first, build on those to “learn to learn”

Page 47: Never Ending Language Learningmausam/courses/col772/spring2016/... · 2016-04-05 · Tenet 1: Understanding requires a belief system We’ll never produce natural language understanding

What next for NELL?

• micro-reading [Krishnamurthy, Betteridge]

• beyond English [Hrushka]

• add computer vision [Gupta, Chen]

• scalable inference over KB [Cohen, Gardner, Talukdar]

• temporal scoping [Wijaya, Talukdar]

• merge with Freebase, Yago [Wijaya, Talukdar]

• goal-driven, targeted reader [Samadi, Kisiel]

• conversational agent on Twitter [Hrushka, Ritter]

• map to brain image data on sentence reading [Rafidi]