65
Information Extraction

Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

InformationExtraction

Page 2: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

Announcements•  Courseevalua+on:pleasefillout• HW4extendedtonoon,12/4•  Thursday:bringyourlaptop!

•  Poetrygenera+on•  Finalreview

•  Finalexam:12/21,finaliscumula+ve• Whattopicswouldyouliketohearaboutagain?

• DanJurafsky’stalk:5pmtoday,CEPSRauditorium.Hopetoseeyouallthere!“DoesthisVehicleBelongtoYou?”

Page 3: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

EnablingConnectionsinaHyperconnectedWorldthroughEmotionAITaniyaMishra,Affectiva•  Date:Wednesday,December6th•  Time:7-8pm•  Loca+on:Room504oftheDianaCenter

•  Abstract:•  Weliveinahyperconnectedworldpoweredbysmartdevices.Abigpartofbuildingconnec+onsisrecognizingandrespondingtoemo+ons.Butoursmartdevicess+lllackthisfundamentalaspectofsocialcommunica+on,renderingourinterac+onswithorthroughthemsuperficialandlimited.Nowimagineifwecouldempowerourdeviceswithintelligencetorecognizehumanemo+ons.Theresultswouldbetransforma+ve,rangingfromempathe+cchatbotstopersonalizeddigitalsignagetosmartcarsthatensurethecomfortandsafetyofpassengersbyrecognizingtheiremo+ons.Emo+onAI—emo+ones+ma+onviaar+ficialintelligence—canmakethispossible.

•  Inthistalk,Iwillpresentwhatemo+onAIis;itsprac+calandconsumerapplica+ons;theprocessofdevelopingalgorithmsandmodelstoes+mateemo+onsfromdifferentmodali+eswithapar+cularfocusonvoice,whichismyareaofspecializa+on.IwillalsobrieflypresentmycareertrajectoryandhowIcametopar+cipateinthisexci+ngfield.

Page 4: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 5: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 6: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

Clari@icationsfromlasttime.

Page 7: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

Clarifica+on:ThisisPhrasebasedMT(top)versusNeuralMT(foriden+fica+onofphrases)(bohom)asIpresentedlast+me.

Choetal2014

Page 8: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

Performance•  Withoutahen+on,LSTMworksquitewellun+lasentencegetslongerthan30words

•  Ahen+ondoesbeher,however,evenwithshortersentences

•  OthertricksinWMT2017:•  Improvementsof1.5–3bluepoints(Edin)•  Layernormaliza+on,deepernetworks(encoderdepthof5,decoderdepthof8)

•  BasePhraseEncodings(BPE)•  THESEAREACTUALLY“BYTEPAIRENCODING”.THEYARETOIDENTIFY

SUB-WORDS.PRODUCEONLYSUBWORDSSEENINTHETRAININGCORPUS.Thisrestric+onreducedvocabulary.

•  Reducedvocabularyimprovesmemoryefficiency•  Data:parallel,back-translated,duplicatedmonolingual

Page 9: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

InformationExtraction• Extrac+onofconcretefactsfromtext

• Nameden++es,rela+ons,events

• Openusedtocreateastructuredknowledgebaseoffacts

Page 10: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

• KathyMcKeown,aprofessorfromColumbiaUniversityinNewYorkCity,tookatrainyesterdaytoWashingtonDC.

Page 11: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

NamedEntities• KathyMcKeownper,aprofessorfromColumbiaUniversityorginNewYorkCityloc,tookatrainyesterdaytoWashingtonDCloc.

Page 12: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

NamedEntities,Relations• KathyMcKeownper,aprofessorfromColumbiaUniversityorginNewYorkCityloc,tookatrainyesterdaytoWashingtonDCloc.

• KathyMcKeownfromColumbia• ColumbiainNewYorkCity

Page 13: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

NamedEntities,Relations,Events• KathyMcKeownper,aprofessorfromColumbiaUniversityorginNewYorkCityloc,tookatrainyesterdaytoWashingtonDCloc.

• KathyMcKeowntookatrain(yesterday)

Page 14: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

EntityDiscoveryandLinking• KathyMcKeown,aprofessorfromColumbiaUniversityinNewYorkCity,tookatrainyesterdaytoWashingtonDC.

Page 15: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

IEforTemplateFillingRelationDetectionGivenasetofdocumentsandadomainofinterest,fillatableofrequiredfields.•Forexample:

Numberofcaraccidentspervehicletypeandnumberofcasual+esintheaccidents.

Page 16: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

Never-EndingLanguageLearner TomMitchell CMU

•  Cancomputerslearntoread?

•  Browsesthewebandahemptstoextractfactsfromhundredsofmillionsofwebpages

•  Ahemptstoimproveitsmethodsandaccuracy

•  Todate,50millioncandidatefactsatdifferentlevelsofconfidence

•  hhp://rtw.ml.cmu.edu/rtw/

Page 17: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

Q:WhenwasGandhiborn?A:October2,1869

Q:WherewasBillClintoneducated?A:GeorgetownUniversityinWashington,D.C.

Q:Whatwastheeduca+onofYassirArafat?A:CivilEngineering

Q:WhatisthereligionofNoamChomsky?A:Jewish

IEforQuestionAnswering

Page 18: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

StateoftheArt(English)

•  NamedEn++es(news)•  Rela+ons(slotfilling)•  Events(nuggets)

F-measure

•  89%•  59%•  63%

Methods:Sequencelabeling(MEMM,CRF),neuralnets,distantlearningFeatures:linguis+cfeatures,similarity,popularity,gazeteers,ontologies,verbtriggers

Page 19: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

WhereHaveYouBeenEntityDiscoveryandLinking?GrowwithDEFT 2006-2011 2012-2017HENGJI,RPIMen+onExtrac+on Human(most) Automa+c

NILClustering None 64methods

ForeignLanguages Chinese(5%-10%lowerthanEnglish)

Systemfor282languages(Chinese/Spanishcomparableto/OutperformEnglish);researchtoward3,000languages

DocumentSize - 500à90,000documents

Genre News,webblog News,DiscussionForum,Webblog,Tweets

En+tyTypes PER,GPE,ORG PER,GPE,ORG,LOC,FAC,hundredsoffine-grainedtypesfortyping

Men+onTypes Nameorallconcepts(most)

Name,Nominal,Pronoun(forBeST)

KB Wikipedia FreebaseàListonly

TrainingData 20,000queries(en+tymen+ons)

500à0documents;unsupervisedlinkingcomparabletosupervisedlinking

#(Good)Papers 62 110(newKBPtrackatACL);6tutorialsattopconferences

SlidefromHengJi

Page 20: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

•  <PERSON>AlexanderMackenzie</PERSON>,(<TIMEX>January28,1822<TIMEX>-<TIMEX>April17,1892</TIMEX>),abuildingcontractorandwriter,wasthesecondPrimeMinisterof<GPE>Canada</GPE>from….

•  Sta`s`calsequencelabelingtechniquescanbeused–similartoPOStagging•  Word-by-wordsequencelabeling•  Exampleoffeatures

•  POStags•  Syntac`ccons`tuents•  Shapefeatures•  Presenceinanameden`tylist

ApproachforNER

Page 21: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

•  Givenacorpusofannotatedrela+onsbetweenen++es,trainaclassifier:•  Abinaryclassifier

•  Givenaspanoftextandtwoen++es->decideifthereisarela+onshipbetweenthesetwoen++es

•  Features•  Typesoftwonameden++es•  Bagofwords•  POSofwordsinbetween

•  Example:•  ArentedSUVwentoutofcontrolonSunday,causingthedeathofsevenpeopleinBrooklyn

•  Rela+on:Type=Accident,VehicleType=SUV,casualty=7,weather=?

SupervisedApproachforrelationdetection

Page 22: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

• Paherns:•  “[CAR_TYPE]wentoutofcontrolon[TIMEX],causingthedeathof[NUM]people”

•  “[PERSON]wasbornin[GPE]”•  “[PERSON]wasgraduatedfrom[FAC]”•  “[PERSON]waskilledby<X>”

•  MatchingTechniques•  Exactmatching

•  ProsandCons?•  Flexiblematching(e.g.,[X]was.*killed.*by[Y])

•  ProsandCons?

PatternMatchingforRelationDetection

Page 23: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 24: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 25: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

• Howcanwecomeupwiththesepaherns?• Manually?

•  Taskanddomain-specific•  Tedious,+meconsuming,notscalable

• Machinelearning,semi-supervisedapproaches

PatternMatching

Page 26: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 27: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

1.Name(s),aliases:2.*DateofBirthorCurrentAge:3.*DateofDeath:4.*PlaceofBirth:5.*PlaceofDeath:6.CauseofDeath:7.Religion(Affilia+ons):8.Knownloca+onsanddates:9.Lastknownaddress:10.Previousdomiciles:11.Ethnicortribalaffilia+ons:12.Immediatefamilymembers13.Na+veLanguagespoken:14.SecondaryLanguagesspoken:15.PhysicalCharacteris+cs16.Passportnumberandcountryofissue:17.Professionalposi+ons:18.Educa+on19.Partyorotherorganiza+onaffilia+ons:20.Publica+ons(+tlesanddates):

Task:Produceabiographyof[person]

Page 28: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

• Toobtainhighprecision,handleeachslotindependentlyusingbootstrappingtolearnIEpaherns.

• Toimprovetherecall,u+lizeabiographicalsentenceclassifier

Biography–twoapproaches

Page 29: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 30: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 31: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 32: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 33: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 34: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 35: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 36: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 37: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 38: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 39: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 40: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 41: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

Bouncingbackandforth• Workedwellforafieldssuchaseduca+on,publica+ons,immediatefamilymembers,party,otherorganiza+onac+vi+es

• Didnotworkwellforotherfieldsincludingreligion,ethnicortribalaffilia+ons,previousdomiciles->toomuchnoise

• Whyisthebouncingideabeherthanusingonlyonecorpus?

Page 42: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 43: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

HowareneuralnetsusedforIE?

Page 44: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

OrganizingknowledgeIt’saversionofChicago–thestandardclassicMacintoshmenufont,withthatdis+nc+vethickdiagonalinthe”N”.

ChicagowasusedbydefaultforMacmenusthroughMacOS7.6,andOS8wasreleasedmid-1997..

ChicagoVIIIwasoneoftheearly70s-eraChicagoalbumstocatchmyear,alongwithChicagoII.

44 SlidefromHengJi

Page 45: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

Cross-documentco-referenceresolutionIt’saversionofChicago–thestandardclassicMacintoshmenufont,withthatdis+nc+vethickdiagonalinthe”N”.

ChicagowasusedbydefaultforMacmenusthroughMacOS7.6,andOS8wasreleasedmid-1997..

ChicagoVIIIwasoneoftheearly70s-eraChicagoalbumstocatchmyear,alongwithChicagoII.

45 SlidefromHengJi

Page 46: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

46

Referenceresolution:(disambiguationtoWikipedia)It’saversionofChicago–thestandardclassicMacintoshmenufont,withthatdis+nc+vethickdiagonalinthe”N”.

ChicagowasusedbydefaultforMacmenusthroughMacOS7.6,andOS8wasreleasedmid-1997..

ChicagoVIIIwasoneoftheearly70s-eraChicagoalbumstocatchmyear,alongwithChicagoII.

SlidefromHengJi

Page 47: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

47

The“Reference”CollectionhasStructureIt’saversionofChicago–thestandardclassicMacintoshmenufont,withthatdis+nc+vethickdiagonalinthe”N”.

ChicagowasusedbydefaultforMacmenusthroughMacOS7.6,andOS8wasreleasedmid-1997..

ChicagoVIIIwasoneoftheearly70s-eraChicagoalbumstocatchmyear,alongwithChicagoII.

Used_In

Is_a Is_a

Succeeded

Released

SlidefromHengJi

Page 48: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

48

AnalysisofInformationNetworksIt’saversionofChicago–thestandardclassicMacintoshmenufont,withthatdis+nc+vethickdiagonalinthe”N”.

ChicagowasusedbydefaultforMacmenusthroughMacOS7.6,andOS8wasreleasedmid-1997..

ChicagoVIIIwasoneoftheearly70s-eraChicagoalbumstocatchmyear,alongwithChicagoII.

SlidefromHengJi

Page 49: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

49

Here–Wikipediaasaknowledgeresource….butwecanuseotherresources

Used_In

Is_a Is_a

Succeeded

Released

SlidefromHengJi

Page 50: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

Wiki@ication:TheReferenceProblem

Blumenthal(D)isacandidatefortheU.S.SenateseatnowheldbyChristopherDodd(D),andhehasheldacommandingleadintheracesinceheenteredit.ButtheTimesreporthasthepotentialtofundamentallyreshapethecontestintheNutmegState.

Blumenthal(D)isacandidatefortheU.S.SenateseatnowheldbyChristopherDodd(D),andhehasheldacommandingleadintheracesinceheenteredit.ButtheTimesreporthasthepotentialtofundamentallyreshapethecontestintheNutmegState.

CyclesofKnowledge:Groundingfor/usingKnowledge

50SlidefromHengJi

Page 51: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

TaskDe@inition• Aformaldefini+onofthetaskconsistsof:

1.  Adefini+onofthemen`ons(concepts,en++es)tohighlight

2.  Determiningthetargetencyclopedic

resource(KB)

3.  DefiningwhattopointtointheKB(`tle)

51SlidefromHengJi

Page 52: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

Blumenthal(D)isacandidatefortheU.S.SenateseatnowheldbyChristopherDodd(D),andhehasheldacommandingleadintheracesinceheenteredit.ButtheTimesreporthasthepoten+altofundamentallyreshapethecontestintheNutmegState.

Blumenthal(D)isacandidatefortheU.S.SenateseatnowheldbyChristopherDodd(D),andhehasheldacommandingleadintheracesinceheenteredit.ButtheTimesreporthasthepoten+altofundamentallyreshapethecontestintheNutmegState.

ExamplesofMentions(1)

52SlidefromHengJi

Page 53: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

NeuralApproachtoEntityLinking(Wiki@ication)Gupta,SinghandRoth,EMNLP2017• Learnsadense,unifiedrepresenta+onofen++es•  Encodesseman+candbackgroundknowledgefrommul+plesources

•  Anencoderforeachsourceofinforma+on•  En+tyembeddingslearnedtobesimilartoencodings

• OnlyusesindirectsupervisionfromWikipedia/Freebase

• Canincorporatenewen++eswithoutretrainingexis+ngrepresenta+ons

• hhp://cogcomp.org/papers/GuptaSiRo17.pdf

Page 54: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

JointlyEmbeddingEntityInformation

Page 55: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

JointlyEmbeddingEntityInformation

Page 56: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

LookatWikipedia• En+tydescrip+on:hhps://en.wikipedia.org/wiki/India_na+onal_cricket_team

Page 57: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

EncodingthementioncontextIn1932,IndiaplayedtheirfirstgameinEngland.• Examplemen+oncontainstwomen+ons:“India”and“England”

• Aimtodisambiguate“India”totheteam•  Localcontext:“played”and“match”• Documentcontext:toiden+fythesport

• Preservetheseman+cs:“England”shouldnotmatchtoateam

Page 58: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

LocalContext• Givenmen+onminsentence:{w1,…,m,….wN}•  LepLSTMappliedtow1…m->

• RightLSTMappliedtom….wN->•  concatenatedandpassedthroughasinglelayerfeedforwardnetwork

Page 59: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

DocumentContextEncoder• Bagofmen+onsvector:

•  USA,PearlJam,NasserHassain• Compressedtoalowdimensionalrepresenta+onusingasinglelayerfeedforwardneuralnetwork

• Combinelocalanddocumentrepresenta+onstogetamen+onlevelencodingusingconcatena+onandfeedthroughasinglelayerfeedforwardnetwork

Page 60: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

EncodingEntityDescriptionD• EmbedeachwordoftheWikipediadescrip+onasad-dimensionalvector

• EncodeasafixedvectorusingaCNN:

Page 61: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

LearningtheTypeRepresentation• EmbedtypeTinFreebase

• Eachen+tycanhavemul+pletypes

• Jointlylearnen+tyandtyperepresenta+ons

Page 62: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

LearningUni@iedEntityRepresentations• Separatemodelsforen+tymen+ons,en+tydescrip+ons,typedescrip+ons

• Tolearnthedifferenten+tyrepresenta+onsandtheirparameters,jointlymaximizethetotalobjec+vewherevearethesetofen+tyrepresenta+onsandθaretheparameters

Page 63: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 64: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822
Page 65: Information Extraction - Columbia Universitykathy/NLP/2017/ClassSlides/... · • Alexander Mackenzie , (January 28, 1822

Lookingforward• Morelanguages:3000!

• Mul+-media

• Streamingmode

• Nomoretrainingdata

• Context-aware,living