论文英语摘要

Abstract

MoreandmorelinkedopendataandusergeneratedcontentsarepublishedontheWebaftertheSemanticWebhasbeenproposed;thewebisnowchangingfromthewebofdocumenttowebofdata,whichcontainsabundantentitiesandrelations.KnowledgegraphhasbeenfirstproposedbyGoogle,whichisfocusedondescribingvariousrealworldentitiesandconcepts,andtheirrelations.Knowledgegraphisanewvisionofontology;aknowledgegraphextendsanontologyintheentitylevel.Ontologyusuallyfocusesonconceptsandtheirrelations,whichspecifiestheschemaofknowledgegraph;andknowledgegraphaddsalargenumberofentitiesintoontology.Knowledgegraphiswidelyusedinsemanticsearch,intelligentquestion-answering,knowledgeengineering,datamining,anddigitallibrary.

Thisthesisstudiesknowledgeconstructionfrommultipledatasourcesbasedonthecurrentresearchachievementsaboutknowledgegraphandontologyconstruction.Themainworksandcontributionsofthisthesisareasfollows,1.Thisthesisexploitsthestructuredandsemi-structureddataonthewebforknowledgeconstruction,suchaslinkedopendata,onlineencyclopedias,anddomainwebsites,whichhavehighcoverageandquickupdatespeed.Thisthesisexplainshowtoextractandlearnknowledgefromthem,andhowtoensurethequalityoftheconstructedknowledgegraph.

2.Westudyonconstructknowledgegraphfrommultipledatasourcesandcombinetheadvantagesofthem,includingthegoodprecisionofthedatainrelationdatabases,thehighcoverageofthelinkedopendataandpublicknowledgebasesontheweb,andthedepthofthedomainorienteddata.Thethesisproposesaknowledgeconstructionmethodfromthesemultipleresources,andensurestheprecisionofthelearntknowledgegraphbasedontheredundanciesofdifferentresources.

3.Thisthesisalsoexplainshowtoexactknowledgefromlarge-scalewebtextandproposesanopenrelationextractionmethodbasedonself-supervisedlearning.Theextractedrelationsincludesynonym,hyponymyandattributerelationsamongconceptsandentities.

Themainadvantageofthemethodisthatitlabelstrainingsamplesautomaticallybyusingknowledgeextractedfromstructureddataorsemi-structureddataandsomegeneralheuristicrules.Inordertoobtaintextautomatically,thisthesisalsoproposesaheuristicrulebasedwebinformationextractionalgorithmtoextractmaincontentofwebpages.

4.Fordomainknowledgegraphconstruction,thisthesisalsofocusesonhowtousethedomainstructureddataanddesignsamappinglanguagewhichspecifieshowtomapdatainrelationdatabasesintoknowledgeinknowledgegraph.Thisthesisalsostudieshowtoautomaticallydiscoverdomaindatasourcesuchasopendomainknowledgebasesandwebsitesontheweb5andproposedacorrespondingalgorithm.

5.Thisthesisalsodevelopsanonlinecollaborativeknowledgegrapheditplatformwhichtriestoleveragecrowd-wisdomforknowledgegraphgeneration.Themainadvantagesoftheplatformareitsabilityforconcurrentediting,anditcouldcombinewiththeautomaticallearningalgorithms.

Finally,weconstructageneralknowledgegraphwith7,392,384entitiesand60,842,064factsbasedontheproposedalgorithms.Whilecomparingtheconstructedknowledgegraphwithotherknowledgedatabasesanddatasets,wefindithasgoodcoverage;meanwhile,theaverageprecisionoftheknowledgeisabove95%.Moreover,weconstructadomainknowledgegraphaboutfisheswhichcontainsmorethan32thousandsfishspecies;itsgoodcoveragecomesfromtheusageofexistingmostcompletedatasources.

Abstract

ThepresentpaperexploresthecategorizationofChineseandEnglishthree-wordidiomsanddiscussesthedifferencesofthesyntacticfunctionsbetweenthoseidiomsofthetwolanguages.Thecategorizationofthree-wordidiomsismotivatedbytheclassificationofidiomsinFillmoreetal.⑴andtheautonomy/dependencealignment[2].Toimprovetheclassification,wealsoborrowtheconceptofchemical-bond.Inthepresentstudy,verbsinthethree-wordidiomsareseenasequaltothemetallicionsinchemistry.BothChineseandEnglishthree-wordidiomscanbedividedintoionic-bondtypeandcovalent-bondtype.

Moreover,basedonthemodeloftheautonomy/dependencealignment,theChinesethree-wordidiomscanbefurtherclassifiedinto11subtypes,whileEnglishthree-wordidiomswillbecategorizedinto9subtypes.

Thecontrastiveresearchofthesyntacticfunctionsisconductedonthebasisoftheclassification.WestartwiththecontrastivestudyofthecategorizationbetweenChineseandEnglishthree-wordidioms.Qualitativestudyisthefirststepinthediscussionofthesyntacticfunctions.Theexamplesareobtainedthroughthecorpusandthesyntacticfunctionsofthethree-wordidiomswillbeanalyzedwithinthesentences.Indeed,wecandrawsomeconclusionsfromthequalitativestudy,butthelackofrepresentativenessandreliabilitymakesthestudyincomplete,callingforththequantitativestudy.Inthepresentpaper,1000Chinesethree-wordidiomsand1000Englishthree-wordidiomsarepickedupfromtheDictionaryofIdiomsfromtheAncientChineseNovelsandNTC'sAmericanIdiomsDictionaryrespectively.

WiththeassistanceofSPSSandcorpus,wehavedonesomedataprocessingonthesyntacticfunctionsofthethree-wordidioms.Aspecialnoteforthisisthatthechi-squaretestinSPSSwillprovethereliabilityandrepresentativenessofthesampletestinthepresentstudy.

Wefind,firstofall,32.3%oftheEnglishcovalent-bondidiomsactasobjectsorsubjectsinthesentencesand67.7%asmodifiersorcomplements;25.7%oftheChinesecovalent-bondidiomsactasobjectsorsubjectsand74.3%asmodifiersorcomplements.

Secondly,alloftheEnglishionic-bondidiomscanfunctionaspredicatesand97.8%ofthemcanalsobemodifiersorcomplements;86.5%ofChineseionic-bondidiomsactaspredicatesand13.5%asmodifiersorcomplements.Accordingtothis,thedifferenceandsimilarityofthesyntacticfunctionsbetweenChineseandEnglishthree-wordidiomscanbeconcludedasfollows:alltheEnglishionic-bondidiomsactaspredicateswhileinChineseover80%oftheionic-bondidiomsfunctionaspredicatesleaving13.5%functioningasmodifiersorcomplements;thesyntacticfunctionsofcovalent-bondidiomsaresimilarinChineseandEnglish:about70%ofthecovalent-bondidiomsfunctionasmodifiersorcomplementswhileabout30%actasobjectsorsubjects.

Thefindingsofthiscontrastivestudywillhelpanalyzethesyntacticfunctionsofthethree-wordidiomsinthesentencesandassistthetranslationofthethree-wordidioms.

Moreover,thepresentstudyisthefirstresearchontheclassificationofthree-wordidioms,which,wehope,willbeofhelpforthestudyofidioms.

Abstract

Golfappareliscategorizedasoutdoorwears.Originatedinthe14thcenturyinScotland,golfwasanaristocraticsportwithalonghistoryanditwasvotedasthecompetitioneventinthe2016OlympicGames.Asasportinheritingthearistocratictraditionandculture,golfapparelalsounderwentdrasticreformsandevolutionsfromatraditionalaristocraticphasetoamodernstyleandfromacomplicatedpatterntoasimpletrend.Golfappareliscategorizedasoutdoorclothingcatalogueinthewesternpartofthedresssystem.Unlikeathleticsportswear,itlikescasualclothingratherthanathleticsportswear.Likehunting,shooting,horseracingandfishingwhichEnglish-speakingcountrygentlemenpursued,golfrepresentsawayoflife.GolfapparelexperiencedachangefromcomplextosimpleandsawthecreationofsomeclassicoutdoorclothingsuchastheNorfolkjackets,casualjacket,Polo-shirt,etc.Thismanuscriptfocusedonanalyzingtheevolutionofgolfapparelfromthe20thcenturytilltoday.Historicaloriginsandperspectiveofthegolfapparelwasdiscussedtofacilitatetheunderstandingofitshistoryandculture.Thisstudyalsoinvestigatedtherevolutionofthegolfapparelinspiredbythemoderntechnologyandinreverse,itslong-termimpactonthedailylifestyles.

Abstract

Withtheriseofinternationalcommunication,learningEnglishhasbecomeanindispensibletaskfacedbytheminoritynationalitiesinChina.Hence,thenumberoftrilingualisrocketing.Amongthem,Mongoliantrilingualisfeaturedwithuniqueness.

Thisarticleinvolvesfourgroupsofcognitiveexperimentfromvisualandauditorychannels,aimingtoexplorethetimecourseofsemanticactivationintheprocessoflexicalaccess,theimpactofcharactersandspeechonlexicalaccessofEnglishandtherelationshipofL3（English）toLI（Mongolian）andL3（English）toL2（Chinese）。

Thedesignisbasedonevent-relatedpotentials（ERPtechnology），whichrecordsthesensoryactivityandcognitiveprocessonlinecontinuously.SeparatingERPcomponentsfromthescalptodeterminethepsychologicalactivityofactivatedbrainarea,wecmlocatetheendogenousnerveofovertbehavior.Thestudyresultsserveasaneffectivesupplementincognitivestudyofvocabulary.Besides,itcanprovidetheoreticalguidanceforEnglishteaching.

Specificconclusionsinclude:1）TheexitsofprimingeffectinallfourgroupsofexperimentindicatethelexicalconceptofMongolian,ChineseandEnglishisstoredtogether.Allthreelanguageshassharedsemanticlayerbutdifferentconceptlayers,andtheconceptualrepresentationofEnglish（L3）isaccessedthroughtranslationofMongolian（LI）andChinese（L2）asmediation;2）Intermsoftheinteractionofthreelanguages,MongoliancollegestudentsarebetteratthejudgmenttaskwhenthematerialpresentedisL3（English）toL2（Chinese）ratherthanL3（English）toLI（Mongolian），whichcanbededucedthatthethirdlanguageacquisition（English）dependslargelyonthesecondlanguage（Chinese）ratherthantheirmothertongue（Mongolian）；3）TheemergenceofP300indicatestheprocessingofspeechtakeplacebeforesemanticprocessingofEnglishvocabulary.Inaddition,thesemanticprocessingtakesshortertimethoughtherepresentationofspeechthancharacters.ThosefactsproveMongoliantrilingualachievesemanticaccesswiththehelpofspeechactivation,whichislargelyaffectedbytheirmothertongue.

Abstract

Emotionalmeaningplaysanimportantroleindiscoursecomprehension.Itisoftheoreticalsignificanceandimportantapplicationvaluetodeeply.revealreaders'specificcognitiveprocessesinmindwhentheycomprehendemotionsindiscourse.

However,traditionalliteraryanalysisregardsemotionsindiscourseasrhetoricaleffectsandresultsofartisticexpressionswithinternalcomprehensionprocessesneglected.

Psychologicallinguisticsandcomputationallinguisticstreatthemasdirectinterpretationorreasoning,equatingpropositionalinformationwithcomprehensionresults.Cognitivepoeticsrevealsthatitisthetangibleemotionalresponsesthataretriggeredbyemotionalcomprehensionindiscourse,whichwilladverselyaffectthecomprehensionprocessingofdiscourse.Inthisprocess,readers'self-conceptandemotionalexperienceplayavitalpart.Butwhatpreviousstudieslackisthespecificdiscussionofinternalprocessesonthesephenomena.

Mentalspacesandconceptualintegrationtheory（MSCIforshort）targetdescribingmidexplainingonlinemeaningconstruction.Recently,researchesonpoetry,novelsandsuper-longtextsbymeansofMSCIareconstantlyemerging,whichhavebeenmakinglotsofcontributionstodiscourseinterpretation.However,asFreeman（2006）pointedoutthatdescribingandexplainingemotionsindiscourseisstillintheblank.ThisdissertationattemptstosolvetwoproblemsbymeansofMSCI:thefirstoneistheonlineprocessinvolvedinthecomprehensionofemotionalmeaning;thesecondoneistheimpactoftheseprocessesonmentalcoherenceindiscourse.

Tobridgethegap,thisstudydefinesemotionasanepisodeonthebasisoftherecentdevelopmentofcognitivescience,andanalyzesthebasicdepartmentsandmentalcomponentsofemotionalprocessandtherelationshipsamongtheseingredientsaccordingtotheepisodemodelofemotions.Then,theon-lineemotionalmeaningconstructionindiscoursecomprehensionisanalyzedwithexamples,whichdistinguishesdifferentemotionalmentalspacesaccordingtotheirmentalmechanism.

Particularly,integrationprocessesinvolvedinon-lineemotionalmeaningconstructionindiscoursecomprehensionareprobed,includingthenatureofcognitiveoperationandvitalrelationssuchasselectivemappings,etc.Furthermore,anexperimentisconductedtoinvestigatetheunderstandingdifferencesundertheconditionofdifferentemotionalprompts.Finallyweanalyzetheconnectingandnestedrelationbetweenemotionalmentalspacesformedthroughon-lineemotionalmeaningconstructionindiscoursecomprehensionandthecorrespondingconceptualintegrationprocesswithexamples.

Morethanthat,thedifferencesarecomparedbetweenmentalcoherenceindiscoursetakenshapebyemotionalmentalspacesandtheirconceptualintegrationprocessesandthestructureformedbytraditionaldiscretefactorsliketime,space,cause-effect.What'smore,thispaperattemptstoanalyzemediumandmacroscopicstructureofdiscoursefromtheperspectiveofemotions.

Thefindingsofthisresearchareconcludedasthefollowing:Firstly,thecomprehensionprocessmustcontainconceptualintegrationprocessofactivatinglong-termmemorytobuildupexperientialspace,andthefusionofexperientialspaceandnarrativespacetoshapeintoblendedspace.Blendedspacedemonstratestheemergentpropertyofemotionalmentalspaceinthecaseofemotionalarousalsexcludedandincluded.“Eventcognition-emotion”relation,“emotion-consequentbehavior”

relationand“emotion-cognitivecolor”relationaretheveryvitalrelationsinthecomprehensionprocessofemotions.Thesevitalrelationsarethecognitivemechanismof“emotion-cognition”interaction.“Emotion-cognition”interactivemechanismisthespecificmechanismofcognitiveoperation（includingcomposition,completion,elaborationandsoon）inemotionalmentalspace.Itmakesemotionalandnarrativedescriptionindiscoursemutuallynurishandpromotetomanifestthegestalteffect（Thewholeisgreaterthanthesumofitsparts.），engenderingrichbeyondliteralmeaningsandtwo-wayvalue-addedpragmaticeffect.

Secondly,Itisdemonstratedthat“eventcognitionemotion”relation,“emotion~consequentbehavior”relation,“emotion-~cognitivecolor”relationandemotionalcontraststructuremakeemotionalinformationindiscourse（oftenconstructedbeyondliterally）andrichnarrativecontentnaturallyjointtogether,andunitecharacters,scenes,events,etc.,inrelatedcontents.Emotionexecutesfunctionsofreference,consistency,governace,reason,andintegrationinthesejoints.Onthemacroscopiclevel,emotionsactivatedinonlinediscourseprocessingbythereadercanimplementfunctionsofcompression,generalization,markandlayout,etc.forinformationindiscoursetoformglobalthematicimpressionofdiscourseandoutlinestructure.Thecoherentrelationshipandstructuregroundonemotionalmentalspacesystemisanewwaytotakeshapecoherentdiscourseandstructuredifferentfromgeneralcognitivefactors,suchastime,space,logic,comparison,analogy.Itmanifeststhedominanceofemotionalprocessingandspecialgestaltmentalrepresentationindiscourseby“emotion-cognition”interactivemechanism.

Ingeneral,thisresearchexplainsspecificemotionalmeaningbytheuseoftheintegratedrelationshipbetweenemotionalarousalsandcognitiveprocessingtobuildupthecompositionalviewofemotionalcomprehension.Itclassifies“eventcognitionemotion”relation,emotion一consequentbehavior,'relation,“emotion~cognitivecolor''relationandemotionalcontraststructureonthestrengthoftheanalysisofemotionalprocesstoformaunifiedtheoreticalframework,basedonwhichweanalyzeon-linemeaningconstructionandcoherentstructureofemotionaldiscourse.Italsopointsoutthenecessityofconceptualintegrationintheprocessofemotionalunderstandingindiscourse,whichmakesemotionalunderstandingembodiedandachievesgestaltcomprehensioneffectthroughinteractionbetweencognitionandemotion.What'smore,itproposesanewperspectiveofdiscoursecoherence,namelyemotionalmentalspaces,toanalyzelocal,mediumandmacroscopicstructureofcoherence,andthendiscoverstheindependentstructuralsystemofimplicity,depth,highorderandintegrationinemotionaldiscourse.Accordingly,theaboveinnovativeworkinthisstudyextendstheapplicationofMSCIandenrichestheexplanationofCognitiveLinguisticstodiscoursestructureandcoherence.What'smore,thisstudyalsoattemptstotestifyMSCIwithempiricalevidences.

Themainlimitationofthisstudyisfocusingonstaticdiscourse.Inaddition,theanalyzingmaterialsareconfinedtoChineseandlackmultilingualcontrastandcomparison.Futureresearchshouldfurtherdiscussinteractivediscourseanalysisandcontrastandcomparisonofcross-culturalcommunication.