Affect, Crea+vity, and Humor: Computa+onal Challenges Alessandro Valitu< University College Dublin Invited Talk at the Department of Computer Science, University of Bari “A. Moro” 25 May 2016
Affect,Crea+vity,andHumor:Computa+onalChallenges
AlessandroValitu<UniversityCollegeDublin
InvitedTalkattheDepartmentofComputerScience,UniversityofBari“A.Moro”25May2016
AbstractThelastdecadehasseentheemergenceofnewresearchareascalled"affec+vecompu+ng","computa+onalcrea+vity",and"computa+onalhumor".Theautomatedrecogni+onofsen+mentsandopinionshasbecomeastabletrackofthetopAr+ficialIntelligenceconferences.Agrowingnumberofinterna+onalprojectsarefocusingontheimplementa+onofformsofcrea+vity.Morerecently,theideaofbuildingcomputerprogramscapableofrecognizingandgenera+nghumorisnotconsideredunachievableasinthepast.Inthistalk,Iwillgiveasummaryofmyexperienceinthesenewareasofcomputerscience.Morespecifically,Iamgoingtodescribesomeoftheideas,resources,andmethodsdevelopedduringmyresearchac+vity.
OutlineAffect:• Affec+velexicalresources:WordNet-Affect• LatentSeman+cAnalysisforaffectrecogni+on:Affec+veWeight• Affec+vetextanima+onCrea+vity:• Computa+onalCrea+vityareas• Crea+veSystems• DynamicalsystemsandbasinjumpingHumor:Computa+onalhumorgenera+onHumorousvaria+onoffamiliarexpressionsLexicalreplacementhumor
WordNet-Affect• WordNetisanon-linelexicalreferencesystemwhosedesignisinspiredbypsycholinguis,ctheoriesofhumanlexicalmemory
• Englishnouns,verbs,adjec+vesandadverbsareorganizedintosynonymsets(synsets),eachrepresen+nganunderlyinglexicalconcept
• InWordNet-Affectwehaveanaddi+onalhierarchyofaffec,vedomainlabels
A-LabelsandsomeexamplesA-Label Examples of Synsets
EMOTION noun "anger#1", verb "fear#1"
MOOD noun "animosity#1", adjective "amiable#1"
TRAIT noun "aggressiveness#1", adjective "competitive#1"
COGNITIVE STATE noun "confusion#2", adjective "dazed#2"
PHYSICAL STATE noun "illness#1", adjective "all_in#1"
HEDONIC SIGNAL noun "hurt#3", noun "suffering#4"
EMOTION-ELICITING SITUATION noun "awkwardness#3", adjective "out_of_danger#1"
EMOTIONAL RESPONSE noun "cold_sweat#1", verb "tremble#2"
BEHAVIOUR noun "offense#1", adjective "inhibited#1"
ATTITUDE noun "intolerance#1", noun "defensive#1"
SENSATION noun "coldness#1", verb "feel#3"
Freely available (for research purposes) athttp://wndomains.itc.it
Valencetagging
• Dis+nguishingsynsetsaccordingtoemo+onalvalence
• Posi,veemo+ons(joy#1,enthusiasm#1),• Nega,veemo+ons(fear#1,horror#1),• Ambiguous,whenthevalencedependsonthecontext(surprise#1),
• Neutral,whenthesynsetisconsideredaffec+vebutnotcharacterizedbyvalence(indifference#1)
Affec+veHierarchy(Screenshots)
Affec+veWeight
1. WordNet-Affectprovidestherepresenta+onofdirectaffec,vewords
2. LatentSeman+cAnalysis(LSA)givesameasureofthesimilaritybetweendirectaffec+vetermsandindirectaffec,vewords
LSAspaced3
d1
d2
synset = w1+ w2 + w3
term = w1
§ Similarity: cosine among vectors
Anexample:universityRelated emotional terms Positive emotional category
university Enthusiasm
professor Sympathy
scholarship Devotion
achievement Encouragement
Related emotional terms Negative emotional category
university Downheartedness
professor Antipathy
study Isolation
scholarship Melancholy
Affec+veweightofnews+tles
• Weanalyzedthewordscontainedineachsentence,andextractthewordwithhigheraffec+veweight
News titles (Google-news) Emotion Word with highest affective weight
Review: `King Kong’ a giant pleasure Joy pleasure#n
Romania: helicopter crash kills four people Fear crash#v
Record sales suffer steep decline Sadness suffer#v
Dead whale in Greenpeace protest Anger protest#v
Textanima+on(1)
• Kine,ctypography:textsthatusemovementsorotherperceptualchangesover+me
• Itaddsafurthercommunica+vedimensiontosimpletext
• Wewanttoexploitalinkbetweenlexicalseman+csoftextsandsomekine+cproper+esforanima+ngthem
Textanima+on(2)
• Weusedasastar+ngpointthekine,ctypographyengine[Leeetal.2002]
• Webuiltadevelopmentenvironmentandascrip,nglanguagefordynamicalcrea+onoftextanima+ons
• Composing(e.g.joining,addinginparallel,…)elementaryanima+onsasbuildingblocks
• Elementaryanima+ons:linear,oscillate,pulse,jitter,etc.
Kine+cbehavior:e.g.“anger”
We annotated each emotion category in Wordnet-affect with an appropriate kinetic behavior
Emo+onalkine+cbehaviors
• E.g.Imita+nghumanresponses
• Joy:asequenceofhops• Fear:palpita+ons• Anger:strongtrembleandblush• Surprise:suddenswellingoftext• Sadness:textdefla+onandsquashing
Affec+veTextAnima+onü Newsheadlines(takenformGoogle-News)
• Twomainsteps:(i)emo+onrecogni+onand(ii)kine+canima+onassembling
1. Recognizetheemo+onalcategoryoftheheadline2. Markthewordsthatareclosertothatemo+on3. Assigntheproperaffec+veanima+ontoeachword4. Assembleacomprehensiveanima+onscript,anddisplaythe
animated+tle
IronyDefini+onTheterm"irony"canrefertodifferentconcepts:• Situa+onalirony:situa+oncharacterizedbycontrastbetween
realityandhumanidealsorinten+ons(alsocalled"ironyofthefate").
• Verbalirony:isarhetoricaldeviceinwhichtheintendedmeaningofstatementsisdifferentfrom(andtypicallyoppositeof)theliteralmeaning.Whatissaidisoppositeofwhatismeant.
• Ifonelooksoutofhiswindowatarainstormandremarkstoafriend,"Gloriousday,isn'tit?"thecontradic+onbetweenthefactsandtheimplieddescrip+onisaformofverbalirony.
• Seman+ccontrastcanbeemployedinbothtypesofirony.Wefocusonverbalirony.
• Polarity-based(verbal)irony:Thepolarityofwhatissaidisoppositetothepolarityofwhatismeant.["glorious"->"miserable”]
Irony
ConceptualCrea+vity Linguis+cCrea+vity
Situa+onalIrony VerbalIrony
Polarityopposi+onasaseman+cdeviceforbothconceptualandlinguis+ccrea+vity
ResearchQues+ons
• Towhatextentisitpossibletogenerateverbalironyautoma+cally?
• Howcanweevaluateverbalironyofcomputer-generatedsentencesusingacrowdsourcingsystem?
• Towhatdegreeseman+ccontrastusedinsitua+onalironyisalsocapabletoachieveverbalirony?
@MetaphorMagnet(1)• ItisaTwioerbot@MetaphorMagnet,whichusesastoreofknowledgetocreatetweetsthataremeanttobemeaningful(Veale2014).
• Thesystemcangeneratearichrangeofcrea+vestatements,whichareregularlypostedonTwioer.
Example:Rememberwhenpeacewasencouragedbynonviolentpeacemakers?Now,peaceisavictoryenjoyedonlybyconqueringvictors.
@MetaphorMagnet(2)
@MetaphorMagnet(3)Asubsetsofthetweetpaoernsaredesignedtobeinten+onallyironic.#Irony:Whensomesovereignslivein"magnificent"palacesthewayrappersliveinwretchedgheoos.#Sovereign=#Rapper#Palace=#Gheoo#Irony:Thinkersformula+ngpromisingideasaboutfu+lefantasy.#Promising=#Fu+le#IdeaAboutFantasy#Irony:Themostdignifiedstatesmanisnotmorecelebratedthanthemostunprofessionalblogger.#Statesman=#Blogger
IronyFactors,IronyMarkers,Echo
Therearethreemainbuildingblocks:1. Seman+cContent:IronyFactors2. Pragma+ccues:IronyMarkers3. Background(i.e.contextual)knowledge:
EchoedInforma+on
#Irony:Thinkersformula+ngpromisingideasaboutfu+lefantasy.#Promising=#Fu+le#IdeaAboutFantasy
IronyFactors,IronyMarkers,Echo
Therearethreemainbuildingblocks:1. Seman+cContent:IronyFactors2. Pragma+ccues:IronyMarkers3. Background(i.e.contextual)knowledge:
EchoedInforma+on
#Irony:Thinkersformula+ngpromisingideasaboutfu+lefantasy.
IronyFactors,IronyMarkers,Echo
Therearethreemainbuildingblocks:1. Seman+cContent:IronyFactors2. Pragma+ccues:IronyMarkers3. Background(i.e.contextual)knowledge:
EchoedInforma+on
#Irony:Thinkersformula+ngpromisingideasaboutfu+lefantasy.
IronyFactors,IronyMarkers,Echo
Therearethreemainbuildingblocks:1. Seman+cContent:IronyFactors2. Pragma+ccues:IronyMarkers3. Background(i.e.contextual)knowledge:
EchoedInforma+on
#Irony:Thinkersformula+ngpromisingideasaboutfu+lefantasy.
IronyFactors,IronyMarkers,Echo
Therearethreemainbuildingblocks:1. Seman+cContent:IronyFactors2. Pragma+ccues:IronyMarkers3. Background(i.e.contextual)knowledge:
EchoedInforma+on
#Irony:Thinkersformula+ngpromisingideasaboutfu+lefantasy. (E.g.,arerapoli+calspeech)
Evalua+onofAutoma+c(Verbal)Irony• Wecanuseaironygeneratorasatestbedforthegenera+onofrandomizedsamples.
• Eachexperimentalse<ngcorrespondstodifferentcombina+onsoflinguis+cfeatures.
• Therefore,wecanusethesystemtostudythecontribu+onofspecificlinguis+cfeaturestothepercep+onofirony.
• Ontheotherhand,ifthelinguis+cfeaturescanbecontrolledbygenera+veparameters,theevalua+onisnotonlyabouttheironyoftextsbutalsoaboutthesystemcapabilitytogenerateironictexts.
• Exampleoflinguis+cproper+esthatcannotbecontrolledbygenera+veparameters:sen+mentofthewholesentence.
ExperimentalSe<ngs
BASE Thevegetablesaremixedinhealthfulsalads.
QUOT Thevegetablesthataremixedin“healthful”salads.
HASH #Irony:Thevegetablesaremixedinhealthfulsalads.
QUOT+COMP Thevegetablesthataremixedin“healthful”saladsaretreatedwithpoisonouspes,cides.
QUOT+HASH #Irony:Thevegetablesthataremixedin“healthful”saladsaretreatedwithpoisonouspes,cides.
Computa+onalCrea+vityAreasResearchArea Crea+veSystem
crea+vitystudies:computa+onalmodellingofhumancrea+vity
humans
machinecrea+vity computerprogram:artefactsgenerator
computer-supportedcrea+vity
computer-program+user
Essen+alProper+es
(Ritchie2007,pp.72-73):
1. Typicality:Howmuchtheartefactisanexampleofagivengenre
2. Novelty:howmuchtheartefactisdissimilartoexis+ngexamplesofitsgenre
3. Quality[=Value]
• Typicalitycanbeusedtoevaluate“weakcrea+vity”• Allthreeproper+escanbeusedtoevaluate“strong
crea+vity”
Creative Systems
(Wiggins2006):• Theconceptualspaceisasetofartefacts(inBoden’sterms,concepts)whichareinsomequasi-syntac+csensedeemedtobeacceptableasexamplesofwhateverisbeingcreated.Implicitly,theconceptualspacemayincludepar+allydefinedartefactstoo.
• Exploratorycrea-vityistheprocessofexploringagivenconceptualspace
39ICCBR-1528September2015
Creative Systems
40ICCBR-1528September2015
(Ritchie2007):
• setofbasicitemsB=datatypes(e.g.stringofwords,arraysofpixels)characterisingtheartefactsproducedbytheprogram
• setofresultsR=setofartefactsproducedbytheprogramwithaspecificsetofse<ngparameters
• inspiringsetI:Theconstruc+onoftheprogramisinfluenced(eitherexplicitlyorimplicitly)bysomesubsetoftheavailablebasicitems.Thissubset,whichwewillcalltheinspiringset,couldbealltherelevantartefactsknowntotheprogramdesigner,oritemswhichtheprogramisdesignedtoreplicate,oraknowledgebaseofknownexampleswhichdrivesthecomputa+onwithintheprogram.
Inspiringset(Ritchie2007,pp.76)
Defini+on:“Theinspiringsetisthesetincluding]alltherelevantartefactsknowntotheprogramdesigner,oritemswhichtheprogramisdesignedtoreplicate,oraknowledgebaseofknownexampleswhichdrivesthecomputa+onwithintheprogram.”
• Whyisitusefultointroducetheinspiringset?Wecouldsaythataprogramismorecrea+veifdoesnotreplicate(orcloselyimitate)theinstanceswhichguideditsdesign.
B
Setofbasicitems
B
I
Inspiringset
B
I
Setofresults
R
B
IR
TypicalartefactsTα,1(R)={x∈R|α≤typ(X)≤1}
Tα,1(R)
B
IR
ValuableartefactsVγ,1(R)={x∈R|γ≤val(X)≤1}
Vγ,1(R)
B
IR
TypicalandvaluableartefactsVγ,1(R)∩Tα,1(R)
Vγ,1(R)∩Tα,1(R)
Crea+veSystemsasDynamicalSystems
• Aconceptualspaceofacrea+vesystemcanbedecomposedinseveralregionscalled“basinsofaorac+ons”,eachassociatedtoaspecifictypeofartefacts.
• Weproposetwoformsofsimilaritybetweenartefacts,accordingtotypicality(t-similarity)andvalue(v-similarity).
48ICCBR-1528September2015
The Fractal Tree
49ICCBR-1528September2015
28September2015 ICCBR-15 50
TheFractalTree
28September2015 ICCBR-15 51
28September2015 ICCBR-15 52
28September2015 ICCBR-15 53
28September2015 ICCBR-15 54
28September2015 ICCBR-15 55
28September2015 ICCBR-15 56
28September2015 ICCBR-15 57
28September2015 ICCBR-15 58
28September2015 ICCBR-15 59
28September2015 ICCBR-15 60
28September2015 ICCBR-15 61
28September2015 ICCBR-15 62
28September2015 ICCBR-15 63
28September2015 ICCBR-15 64
28September2015 ICCBR-15 65
28September2015 ICCBR-15 66
28September2015 ICCBR-15 67
28September2015 ICCBR-15 68
28September2015 ICCBR-15 69
28September2015 ICCBR-15 70
28September2015 ICCBR-15 71
28September2015 ICCBR-15 72
Paths and Attractors
Whatisapathinthe“squareoffractaltrees”?1. Asearchintheconceptualspaceofacrea+vesystem2. Atrajectoryinthephasespaceofadynamical
systemAnaQractorisasetofstates(i.e.,elementsofthestatespaceofadynamicalsystem)towardswhichasetofdynamicalpathstendtoevolve.Thesetofdynamicalpathspoin+ngtotheaoractoriscalledbasinofaQrac+on.
73ICCBR-1528September2015
Basin Jumping and Creativity
• Theconceptualspacecanbedecomposedinbasinsofaorac+on.• Star+ngfromasetofpastexamples(inspira+onset),acrea+ve
systemcanreachfromwhichaspecificbasinofaorac+oncanbeexplored.
• “ClassicCBR”allowsacrea+vesystemtoexplorethebasinofaorac+oncontainingthesetofpastexamples(assumedtobetheinspira+onset)
• “Crea+veCBR”shouldallowsacrea+vesystemtoreachbasinofaorac+onsnotcontainingthepastexamples
• Insummary,ifweassumethecrea+vityasasearchintheconceptualspace,ahigherdegreeofcrea+vityisassociatedtothesearchofnewbasinsofaorac+on(basinjumping).
74ICCBR-1528September2015
Double Similarity
• CBRisbasedonthecapabilitytorecognizeanewcaseassimilartooneofthepastcases
• Crea+veCBRshouldemploytwoformsofsimilarity:– T-Similarity:degreetowhichtwoartefactsarerecognisedasbelongingtothesametype
– V-Similarity:degreetowhichthevalueoftwoartefactsisrecognisedassimilar
• T-similarityallowsthesystemtoexplorethecurrentbasinofaorac+on
• V-similarityallowsthesystemtoreachdifferentbasinsofaorac+on
75ICCBR-1528September2015
Humour
• Defini+on:Capabilitytointen+onallyinduce– mirth– laughter– smile
• Mirthinduc+onisperformedthroughlanguage
emotion
cognition
language
(VerbalInten+onal)Humor
HumorinComputers(2)
Computa+onalHumor
• Itisaresearcharealyingattheintersec+onbetweenComputa+onalLinguis+csandAr+ficialIntelligence.
• Itmainlyaimsattheimplementa+onoftoolsabletogenerateandrecognizehumorautoma+cally,butalsocancontributetothestudyofhumorthroughcomputa+onalmodels.
• Prototypesdevelopedsofarareverylimited,ifcomparedtothehumancapabilitytounderstandandproducehumor.Inmostcases,theyareabletoproducespecifictypesofhumoroustexts(e.g.shortjokesorpunsbasedonwordplays).
Computa+onalHumorGenera+onExamplesofoutputsfromhumorgenerators:• Ritchie,Binsted
o linguis+canalysisofjokes[2004]o punningriddle[BinstedandRitchie,1994]o e.g.:“Whatdoyoucallamurdererwithfibre?Acerealkiller.”
• Stock,Strapparavao largescalelexicalresources,seman+crela+ons,crea+vevaria+onso funnyhacronyms[2003]o e.g.:“MIT=MythicalIns+tuteofTheology”
• Nijholt,Tinholto humorousagents:interac+on,appropriatedness,humorousactso anaphoricalpuns[2004]o e.g.:“MaryaskedSusanaques+on,andshegavetheanswer”“Did
Marygivetheanswer?”
FamiliarExpressionsVaria+on
• Wordsubs+tu+ononfamiliarexpressions(e.g.,proverbs,movie+tles,etc.)
• Twowordsgivenasinput:atopicword(e.g.‘holidays’,‘food’,etc.)andanemo+onword(e.g.‘joy’,‘fear’)
• Searchofpossiblecandidatewordsforthesubs+tu+on(inalistofEnglishwords):o wordphone+callysimilartosomewordinatleastone
familiarexpressiono wordseman+callyrelatedtothetopicwordandthe
emo+onword
(Valitu<etal.2008):
Examples
(‘crash’,‘fear)SaturdayFrightFever
(‘surgery’,‘fear’)BacktotheSuture
(‘den+st’,‘fear’)FatalExtrac+on
(‘beach’,‘joy’)TomorrowisanotherBay
StepbyStepExample
surgery
BacktotheSuture
futuresuturejurassicthoracic
aorac+onextrac+on
symptomtherapy
metabolismanalgesicextrac+onthoracicsuture
futureJurassic
abstrac+onaorac+oncontrac+ondiffrac+ondistrac+oninac+onreac+onretrac+onsubtrac+ontransac+on
fear
MovieTitlesandTabooWords
JAPE:Genera+onofHumorousPunningRiddlesExamplebyGraemeRitchie(2004)
Choosetwowordsthatsoundsthesame
What do you call a ?
! A
(Anadjec+ve) (Anoun)
meansroughlythesameas
meansroughlythesameas
Choosetwowordsthatsoundsthesame
What do you call a ?
! A
meansroughlythesameas
meansroughlythesameas
bare bear
naked teddy
bare bear
JAPE:Genera+onofHumorousPunningRiddlesExamplebyGraemeRitchie(2004)
Choosetwowordsthatsoundsthesame
What do you call a ?
! A
meansroughlythesameas
meansroughlythesameas
bizarre
bazaar
market strange
bizarre
bazaar
JAPE:Genera+onofHumorousPunningRiddlesExamplebyGraemeRitchie(2004)
Humorgenera+onbylexicalreplacement
(Valitu<etal.2013)
• ResearchQues+on:towhatextentacomputerprogramcantransformagenericshorttextasinputandmakeithumorouswithasinglewordsubs+tu+on?
• Keyelements:
o automatedgenera+onofhumorousexpressionsbylexicalreplacement.
o AdultHumor
o Useoftaboowordsasawaytointroducefunnyinappropriateness
ExamplesfromDamnYouAutocorrect.com
Correooreautoma+co
Humorgenera+onprocedureHumorgenera+onbylexicalreplacement:1. TheproceduregetsasinputasegmentofEnglishtext
(e.g.:“Leteverythingturnwellinyourlife!”).2. Thenitperformsasinglewordsubs+tu+on(e.g:‘life’
replacedwith‘wife’)andreturnstheresul+ngtext.3. Tomakeitfunny,thewordsubs+tu+onisperformed
accordingtoanumberoflexicalconstraints:formconstraints,tabooconstraint,andcontextualconstraints.
4. Addi+onally,thetextcanbeappendedwithaphrasesuchas“Imean‘life’not‘wife’}”.
FormalconstraintsTheoriginalwordanditssubs+tutearesimilarinform:• Orthographicsimilarity:twowordsareconsidered
orthographicallysimilarifonewordisobtainedwithasinglecharacterdele+on,addi+on,orreplacementfromtheotherone
• Phone+csimilarity:Wecalltwowordsphone+callysimilariftheirphone+ctranscrip+onisorthographicallysimilaraccordingtotheabovedefini+on
• Rhyme:Twowordsrhymeiftheyhavesameposi+onsoftonicaccent,andiftheyarephone+callyiden+calfromthemoststressedsyllabletotheendoftheword
TabooconstraintsThesubs+tutewordisatabooword:
• Taboowordsarewordsnormallyavoidedineverydayconversa+onsbecauseperceivedasinappropriate
• Theycanrefertobodyproducts,bodyparts,sexualacts,ethnicorracialinsults,profanity,vulgarity,etc.
• Wecollectedalistof700taboowordsasunionofthreesubsets:
o Wordsmanuallyselectedfromanavailablelistofwordsreferringtotheseman+cdomain“sexuality”
o Profani+esandslangterms
o Wordsfromexamplesofhumorousautocorrect
ContextualconstraintsThesubs+tu+ontakesplaceattheendofthetext,andina
locallycoherentmanner• Localcoherence=thesubs+tutewordformsafeasible
phrasewithitsimmediatepredecessor,inordertoexpandthetaboomeaningtothephraselevel
• Theseman+ccontrastisassumedstrongerifthetaboowordcomesasasurpriseintheendofaseeminglyinnocenttext.
• Localcoherenceisimplementedusingn-grams.Weusedthe2012GoogleBooksn-gramscollec+on
ExamplesofoutputsExperimentalcondi+on Output
FORM Ohoh...Denmuzchangeplatliao...Gobackhaveyanjiuagain...Not'plat'...'plan'.
FORM Josaskifuwanameltup?'meet'not'melt'!
FORM+TABOO Gotcaughtintherain.Waitedhalfnhourinthebussstop.Not'buss'...'bus'!
TABOO Heypple...$700or$900for5nights...Excellentmasturba+onwifbreakfasthamper!!!SorryImean'loca+on'
FORM+TABOO+CONT Nope...Juzofffromberk...SorryImean'work'
FORM+TABOO+CONT I'vesentyoumyfart...Imean'part'not'fart'...
ConclusiveKeywords• NLrecogni+onvs.NLgenera+on• Interdisciplinarity• Lexicalresourcesandlexicalvectorspaces• Conceptualspacesandexploratorycrea+vity• Evalua+onofnoveltasksandhypothesistes+ng• Focusonsubjec+veresponse• Crea+vereuse• Interac+vity• Mul+modality
LexicalResources
LexicalVectorSpaces
ConceptualSpaces
SpacesofSubjec+veResponse
ReferencesAffec+veNLP:• A.Valitu<,C.Strapparava,andO.Stock.“DevelopingAffec+veLexicalResources”.PublishedinPsychNology
Journal,ISSN1720-7525,2(1):61-83,2004.• C.StrapparavaandA.Valitu<.“WordNet-Affect:anAffec+veExtensionofWordNet”.AcceptedattheFourth
Interna+onalConferenceonLanguageResourcesandEvalua+on(LREC2004).May26-28,Lisbon,2004.• A.Valitu<andC.Strapparava.“InterfacingWordNet-AffectwithOCCmodelofemo+ons”.AcceptedattheThird
Interna+onalWorkshoponEMOTION2010-CorporaforresearchonEmo+onandAffect,pp.16-19,May23,2010,Valleoa,Malta.
• A.Valitu<andT.Veale.InducinganIronicEffectinAutomatedTweets.Acceptedatthe6thInterna+onalConferenceonAffec+veCompu+ngandIntelligentInterac+on(ACII2015),Xi'an,China,pp.153-159,September21-24,2015.
Computa+onalCrea+vity:• A.Valitu<,C.Strapparava,andO.Stock.“TextualAffectSensingforComputa+onalAdver+sing”.Acceptedatthe
AAAISpringSymposiumonCrea+veIntelligentSystems,March26-28,2008,StanfordUniversity,PaloAlto,California.
• AlessandroValitu<.Crea+veSystemsasDynamicalSystems.AcceptedattheWorkshoponExperienceandCrea+vity(ICCBR-2015),pp.146-150,FrankfurtamMain,Germany,28September2015.
Computa+onalHumor:• A.Valitu<.“HowManyJokesareReallyFunny?TowardsaNewApproachtotheEvalua+onofComputa+onal
HumourGenerators”.8thInterna+onalWorkshoponNaturalLanguageProcessingandCogni+veScience(NLPCS2011),2011..
• A.Valitu<,H.Toivonen,A.Doucet,andJ.M.Toivanen."LetEverythingTurnWellinYourWife":Genera+onofAdultHumorUsingLexicalConstraints.ACL2013.
• A.Valitu<,A.Doucet,J.M.Toivanen,andH.Toivonen(2015).Computa+onalGenera+onandDissec+onofLexicalReplacementHumor.NaturalLanguageEngineering.
• AValitu<andT.Veale.InfusingHumorinUnexpectedEvents.HCIInterna+onal2016,Toronto.