descriptive网易轻博客

TheExplorationofUserKnowledgeArchitectureBasedonMiningUserGeneratedContents–AnApplicationCaseofPhoto-SharingWebsiteNanLiang,JiamingZhong,DiWang,andLiqunZhang(&)InstituteofDesignManagement,S.
J.
T.
U.
,Shanghai,Chinazhanglq@sjtu.
edu.
cnAbstract.
Traditionalmethodstoobtainuserneeds,suchasinterview,haveexposedtheincreasinglyseriousproblemofbiasandinefciencywhenmeetingthebloomingofusers.
Thisresearchtriedtoamelioratethesituationbymininguser-generateddataandconstructingcorrespondinguserknowledgesystemswiththehelpofmoderntechnologies.
Withaphoto-sharingwebsiteasastudycase,severaltechniqueshavebeenimplemented,includingimagefeatureextraction,contentanalysisandstatisticalcalculation,toanalyzeusers'char-acteristicsandpreferences.
Theresultsindicatedthatmanyofthesetechniquesarepracticalandeffectiveforfutureresearchinuserexperiencedesign.
Itisforeseeablethatthedomainofthisresearchcanbeexpandedtotextandvoicetoconstructasynthesisapproachforultimatelyunderstandingusers.
Keywords:ImageContentanalysisUserknowledgeExperiencePhotosharingsite1IntroductionInviewoftheconsiderableimprovementofmateriallivingstandardinrecentyears,designersbegintopaymoreattentiontoemotionalandspiritualelementsintheirproductsandservices.
Themajorconsiderationofuserexperiencedesign,orUED,istocreatesatisfying,aestheticandinnovativeproductswhichconstantlymeetuser'sneedsandevenleadthetrendofmodernlifestyle.
Therefore,itisimportantfordesignerstounderstanduserneedsandfurthertranslatethemintoappropriateproducts.
IntheageoftheInternet,thepresenceofblogs,forums,wiki,SNSandRSScombiningwithnewlydevelopedtheoriessuchasSixDegreesofSeparationandtheLongTail,hasmadeuserknowledgeintoanopen,complexandadaptivesystem.
Inthecurrentwebenvironment,thereisanincreasingdiversityintherepresentingformsofuserknowledge,whileusersusuallyfeeleasytoaccommodatethissituation.
Theproblemislefttodesignersonbothacquiringuserknowledgeandconstructingcorrespondingsystems.
Thekeyofuserresearchisminingtheneedsburieddeeplyinusers'mindthroughtheirlanguageanddailybehavior.
Traditionalmethods,includingquestionnaire,interview,observation,focusgroupandpersona,achievethegoalthroughbehaviorSpringerInternationalPublishingSwitzerland2016A.
Marcus(Ed.
):DUXU2016,PartIII,LNCS9748,pp.
180–192,2016.
DOI:10.
1007/978-3-319-40406-6_17observationandcarefullydesignedconversation.
Designersarerequiredtohaveempathyandanopenmindthroughouttheprocess.
Otherwise,badexpressionsmayleadtodifferentorevenoppositeanswers,deviatingfromuser'sreality.
Tocertainextent,traditionalmethodsrevealuserneeds,butsufferfrompoorefciencyandnon-negligibleinuenceofmoodandenvironment.
Hence,theyarenotsuitableforresearchingonmassiveusers.
Ontheotherhand,theoriginalknowledgeproducedbyusersthemselvesbetterexpressestheirrealthought.
Bigdatatechnologyhasmadeitpossibleandcheapertostudylargegroupsofusers.
Tillnow,itisfre-quentlyusedinmanyeldslikenance,onlinebusiness,healthcare,socialsecurityandsmartcity,comparativelyrareinthatofdesign.
Dataminingcanbeanewaspectforextendingthestudyofuserexperienceanduserknowledge.
Thispaperdescribeshowtodigforuserknowledgeandunderstandtheirneedsbylarge-scaledatasearchingandimagecontentanalysistechnologiesandnallyconstructuserknowledgesystemwhichensuresexcellentuserexperience.
Themethodsdescribedinthispaperarealsogoodreferencestootherdesignresearch.
2MethodologyDescription2.
1OverviewThispapermainlyelucidatehowweapplyimagefeaturerecognitionandcontentanalysistechnologiestoobtainresearchvariables,whicharelaterestimatedbysta-tisticalcalculation,inordertoacquireuserknowledgeandconstructcorrespondingsystem.
Thedetailedresearchprocessisasfollows:HowtoacquireuserknowledgeWhenusingcertainproductsorservices,userswouldexchangeinformation(namelywords,imagesandvoice)andthisinforma-tioncouldberecognizedas"userknowledge"sincetheydirectlyreectusers'demands.
Forinstance,usersofphotosharingsocialwebsitesinteractwitheachotherbyuploadingimages,clicking"like",commentingandreposting.
Intheprocessofthistypeofinteractions,usersundoubtedlyleave"internetfootprints"asapartofuserknowledge,whichmanifesttheirattentionandpreference.
Howtoacquireusers'footprintsInshort,onecouldapplyrespectivetechniquestogureoutthefootprintsleftbyusers.
Forexample,equippedwithpublicpro-gramminginterfacesexposedbyrelevantwebsites(e.
g.
WeChatAPI)andwebcrawlerprograms,oneisabletogetusers'informationsuchasimages,texts,andvoice,undercertainagreementofprivacy.
Theemergingofnewtechnologiesfulllsthepurposeofimageanalysis,broadeningtheareaofinformationcaptureandanalysis.
Analysismethodologyandtools.
Threemainmethodshavebeenexploited,includingimagefeatureidentication,contentanalysisandstatisticalcalculation.
2.
2DetailsofThreeMethodsImageFeatureRecognition.
Threeparticulartoolsfallintothiscategory.
TheExplorationofUserKnowledgeArchitecture181Analyzingtoolsforcolorspatialdistribution.
BasedonpixelRGBvaluesofsampleimages,thistoolgeneratescolorspatialpointsandconductsclusteringanddimension-reductionprocessingthroughvectorcalculationandprincipalcomponentanalysis.
Theresultcanhelpresearchersanalyzevariationincolorcharacteristicsofsamplesfromdifferentusers.
Extractingtoolsforsampledominantcolortone.
Basedonthecalculationofpixelcolorfeatures,thistoolrespectivelygeneratestheentirecolorconstitution,bywhichthedominant80%colorsofrawsamplescanberepresented(Fig.
2).
Afterthat,itwillconductbatchprocessinganalysisandgenerateaformforeachsample,manifestingitsdominantcolortoneforfollowinganalysisofmulti-dimensionalcolordeviation(Fig.
1).
Analyzingtoolsforthesimilarityofsampledominantcolortone.
Dependingonsampledominantcolortonedata,thistoolcalculatesthedominantcolortonesimilaritybetweeneachpairamong574samplesandgeneratescsvformatlesastheinputofstatisticalcalculationsinMDSanalysis.
ContentAnalyzingTechnology.
Contentanalysisisatechnologywhichanalyzesthecontentofsamplesandgeneratesastructuredvariablesystemtodescribethesesamplesbymeansoftags.
Thetagsdemonstratethecategoryandorderdescriptionofthesamples,inordertosupportfuturestatisticalanalysisandsearchforsimilarityordifferences.
Fig.
1.
Extractingtoolsforsampledominantcolor(Colorgureonline)Fig.
2.
Analyzingtoolsforthesimilarityofsampledominantcolors(themeDistComputingTool_v1).
182N.
Liangetal.
Baseontheoverallanalysisofsamples,severaldescriptivevariableshavebeenproposedandlabeled.
Inthescopeofthisresearch,alllabelsfallintooneofthefollowingsixcategories:picturetype,picturetheme,composition,meansofexpres-sion,lightandshade,imagestyle.
Next,weintroducethenotionofmatrixofmetricaldatawhichisbydenitionatableformanagingsamplesandcorrespondingvariablelabels.
Allassignmentofvaluestovariablesresultsfromcombinationofimagefeatureandarticiallabeling.
Basedonthismatrix,alldataisimportedintoSPSSafternecessarynormalizationfornextdescriptivestatisticalanalysisandadvancedcalculation.
StatisticalCalculation.
Statisticalcalculationprovidesawaytodiscovertheinternalrelationbetweenobjectiveelementsshownbypicturesandsubjectiverecognitionofusers,bymeansofclustering,multi-dimensionalanalysisandsomeothertools.
Correspondenceanalysisisthemainstatisticalmethodusedinthisresearch.
Theconnectionsbetweenvariablesarerepresentedgraphicallybyinteractionsummarytable.
Thisanalysistechniqueissuitableforsituationswithmanyqualitativevariablesinwhichconnectionsbetweenthesevariablesofdifferentcategoriesistobeestab-lished.
SPSSisaprevalentsoftwareforthiskindofanalysis.
Nowadayscorrespondenceanalysisiswidelyusedinearly-stageconceptdesign-ing,inareasofdevelopingnewproduct,marketpositioningandadvertisement.
Ithasbecomeanimportanttoolfordesignersandmarketresearcherstosolvetheproblemofevaluatingproductproperty,competitorandtargetingmarket.
3CaseStudyofPhotoSharingWebsitesBenetedfrommassivedataminingtechnology,weselectedapopularusecasetolaunchourstudywhichconcentratedonconstructinguserknowledgeofphotosharingwebsitesandfurtheranalyzingtheneedsandpsychologicalfeaturesoftheiractiveusers.
Manyuseractionscanberegardedastheprocessofproducinguserknowledge,includinguploadingphotosandsocialoperationssuchasclickingalike,commentingandreposting.
Inthisscenario,userknowledgeliesintheimages,textanduseractions.
Althoughtextusuallyindicatestheexactthoughtofusers,understandingthemeaningbyprogrammingisveryhardandmostimportantlytextcannotreecttherelationbetweentheimageitselfandusers'judgementonit.
Aftercarefulconsideration,thepopularimagesinphoto-sharingwebsiteswerechosenasthemainobjectforstudying,fulllingthepurposeofmininginformationapropostoimagesitself,userpreferencesandtheirrelation.
3.
1SelectingTargetWebsiteTherearemanywell-knownphoto-sharingwebsitesincludingInstagram,LofterandFlickrbyYahoo.
WenallychoseFlickraftercomparingthefoundationdate,numberofusersandsomeotheraspects.
FlickrisanimagehostingandvideohostingwebsiteandthewebservicessuitewascreatedbyLudicorpin2004,acquiredbyYahooinTheExplorationofUserKnowledgeArchitecture1832005.
Itofferspreeminentservicesincludingpictureuploadingandstoring,classi-cation,taggingandsearching.
Usersneedtollintheirprolesafterregistrationandtheprolescanhelpusinfuturestudy.
Intheuploadingprocess,usersarerequiredtogivethepictureatitle,adescriptionandsometags.
Formanagingphotosmoreeffectively,userscancreate"set",whichissimilartoaphotoalbum.
UsersofFlickrhavevari-ousbackground,fromprofes-sionalphotographerstoPSamateur.
Allofthemenjoyuploadingtheirfavoritepho-tos,addingtagsanddescrip-tionsandcreatingsetsforthem.
Socialoperationsareevenmorepopularsinceeverybodylovesdiscoveringbeautifulpicturesandgrab-bingattentionofothersreectedbythenumberoflikeandcomments.
Thefeatureofaparticularusercanberevealedbythepicturess/helikesandhottestpicturesmanifesttheinclinationofmostusers.
Asaresult,thesehottestpicturesprovideusaneffectivewayofgettingthefeatureswearestudying,analyzinguserdispositionandnallyconstructuserknowledgesystemofthewebsite.
ThepurposeofthisstudyisexploringthetypeandfeaturesofpopularpicturessharedbyFlickrusersanddescribingtheirbehaviorsinFlickr(Fig.
3).
3.
2ProcessofResearchFlickrholdsanannualshownamed"bestshot",selectingthemostpopularpicturesofthatyear.
Weselectedpicturesfrom"2015bestshot"tonarrowdownthesampledomain.
Totally574pictureswerelteredoutthroughourcrawlerprogramsbecausetheyreceivemorethan99commentsorlikes.
Basedonpreviousstate-of-the-artstudies,wedividedalllabelsinto6categories.
Picturetype:daily;documentary;blackandwhite;art;portrait;landscape;abstract;report;Picturetheme:naturalscenery;animalsandinsects;owersandplants;still-lifeobjects;characterportrait;culturalconstruction;sceneofstories;lightrhythm;Composition:nine-squared;diagonal;symmetry;frame;guideline;dynamic;tri-angle;photographicsubtraction;specialangle;repetition;vertical;curve;slash;centripetal;change;S-shape;opentype;balance;Meansofexpression:simplication;choice;comparison;contrast;scenerydepth;background;lines;balance;motion;perspective;reection;Fig.
3.
Flickrwebsite184N.
Liangetal.
Lightandshade:backlight;softlight;capturelight;appropriateexposure;contrastofexposurelevel;lowanglelightsource;regionalexposure;multicolorcontrast;Imagestyle:traditionalnostalgic,romantic,solemnandelegant,deepandsolemn,easydial,decorativearts,comparisonofcool&warm,openmagic,scarceunique,novelandcreative,humansensations,rhythm,non-mainstreamInordertosynthesizetaginformation,thematrixshouldbetransformedintoquestionnaire.
Someexpertsinbothdesignandphotographyassignedthetagsshownabovetothe574samplesbasedoncertainprinciplesexploredinpreviousstudies.
Withthe574samplesandtheirtags,thematrixofmetricaldatawasestablished,ameasuremethodpreviouslymentioned.
ThematrixwasbeingimportedtoSPSSlatter(Fig.
4).
4Result4.
1ResultEvaluationofImageFeatureIdenticationAccordingtothedesignofresearchpreviouslydescribed,theresearchofimagefea-turesmainlyinvolvesfeatureextractionofthesamples.
Theextractionjobincludes:Makequantitativeanalysisbasedoncolorattributesofthesample(samplepixelRGBvalue).
Themainresearchstepsincludeextractingthedominantcolortone.
Accordingtothespecicfeaturesofsamples,thecompositionofthepictureusuallydiffersinmanyways.
Someofthempossessaconspicuousdominantcolortonewhileothersarecomposedofmanycolors.
Whatever,thenumberofdominantcolortonesofcertainsampleisabletorepresent80%ofitscolorinformation.
Therepresentativecolortoneofsamplesisevolvedfromalldominantcolortones,whichisusedtoanalyzesimilaritybetweensamples.
Fig.
4.
MatrixofmetricaldataTheExplorationofUserKnowledgeArchitecture185Thedistancebetweenthecolortones,whichoccupiesrelativelylargerproportionofdominantcolortones,iscalculatedbasedonthecompositionofeachsample.
Figure5illustratethesimilarityofthepositioningofcolorspace,basedonourcal-culationandanalysis.
Figure6illustratethesimilarityanalysisofdominantcolortones,bytheMDSmulti-dimensionalscalingfunctionofthemeDistComputingTool_v1.
InFig.
7,itisobviousthatallofthesam-plesshowsremarkablepatternsonpositioningdistributionofdominantcolortonesimilarity.
Basedonthedistributionofscatteredplots,atwoelementregressionequationisobtainedbytwoordercurvetting:y0:20:27x0:53x2Tomakethedistributionpatternoftheresultmoreeasilydetermined,researcherssupplementinformationforFig.
8and574dominantcolortonepalettewhicharealsopositionedtothecorrespondingscatteredpositions.
Wefoundthatdespitethedifferencesinpropertiesandcontentamongthe574samples,asignicantpatternexistsinthefeaturesofvisualcognitionofdominantcolortones.
Thepatternwasrepresentedbythemildgradientofbrightnessfromdarknessonthelefttobrightnessontheright.
However,noobviouspatternwasrecognizedinverticaldimension.
Inaddition,thesignicanceofsaturationincenterandcenter-rightareasintheU-shapecurveareaishigherthanthatinotherareas.
Fig.
5.
Thesimilarityofthepositioningofcolorspace.
(Colorgureonline)Fig.
6.
ThemeColorPosition-1.
Fig.
7.
ThemeColorPosition-2(Colorgureonline)186N.
Liangetal.
Tosumup,itisconvincingthatthe574samplesprimarilyreectsdifferencesinsaturationandcolortemperatureintermsofcolorproperties,basedontheresultofcolorspacepositioninganalysisanddominantcolortonesimilarityMDSanalysis.
4.
2ResultEvaluationofStatisticalCalculationRecallpreviousdiscussion,correspondenceanalysisisthemainmethodinthisresearch.
Thelocationmapanalysis,resultingfrom574samplesinalldimensions,isdiscussedbelow.
Amongallthedimensions,abundanceofcolortonesisparticularinterestingsothattherstpartofthissectionmakesacomparisonbetweenitandotherdimensionswhilethesecondpartdiscussesresultswithintheotherdimensions.
AbundanceofColorTonesComparetoOtherDimensionsPictureTheme.
PictureThemeThesigvalueis1.
000a,whichindicatesthatthere'snosignicantrelationbetweenpicturethemeandtoneabundance.
Notypicalpatternisrecognizedinthedistributionofthesamplefromdifferenttopics.
Inaddition,thethemeofstilllifeobjectsisrareinthesample.
Composition.
Thesigvalueis1.
000a,onecanseethatmosttypesofthecompo-sitionisinarelativelyconcentratedmannerwhilethediagonaltypeandcurvestypearerelativelyrare(Fig.
9).
MeansofExpression.
Inthisgure,exceptingthelinetype,theperformanceissimilarinthemajorityofthesample(Fig.
10).
LightandShade.
Thesigvalueis1.
000a.
Thereisnoobviouscorrelationbetweenlightingandtoneabundanceinthisdimension.
Meanwhile,lowanglelightsourceismoreuniqueduetothespecialangle(Fig.
11).
ImageStyle.
Thesigvalueis1.
000a.
Imagestyleandtoneabundancehavenosignicantcorrelation.
However,therhythmisrelativelyrare(Fig.
12).
ResultsWithinOtherDimensions.
Overall,threecommonfeatureswerefoundthroughall574samples.
Firstly,intermsofthetype,picturesaboutsceneryordailylivesrankedthehighest;thenfollowsart,documentaryandportrait;reportandabstracthadtheleastquantity.
Secondly,forthecomposition,mostsampleswereshowedinaFig.
8.
PicturethemeFig.
9.
CompositionTheExplorationofUserKnowledgeArchitecture187wayofnine-squaredorsymmetry,whichisassociatedwithhumanaestheticphysio-logicalcharacteristics.
Peoplelikepictureswhichareconciselycomposedwithacer-tainguidanceorrestriction,suchasradialline,leadingline,diagonal,orframe.
Thethirdcommonfeatureliesinimagestyle.
Themostpopularpicturesareusuallyuniqueandrelaxing.
Nostalgic,romantic,solemn,aestheticandnovelingredientsarewelcomeaswell.
Incontrast,popularpicturesarescarcelyinthemesofrhythm,contrastorhumanity.
Thefourresultsofspecicanalysisareshowninfollowinggures.
PictureTypeComparetoImageStyle.
Thecorrespondenceanalysisofpicturetypeandstyles,with574effectivesamplesandSigvaluezero,indicatingthatthereisasig-nicantcorrelationbetweenthetypeandthestyle.
Thecommonaesthetictasteofinclin-ingsceneryanddailytypeofpictureswasverylikelybeingdevelopedalongwiththeevolutionofhumanbeings.
Analysisofthistypeindicatesthatancientprairiescenery,composedbyfreshgrass,lowjunglesandwindingstreams,givescom-fortableandcongruentfeelingstopeoplelivinginnearlyallplaces.
Peopleoftenndsensesofidentityfromdocumentaryandportraitpaintings,makingitthesec-ondpopulartype.
Abstractpicturesareonlyappreciatedbyasmallgroupofpeople(Fig.
13).
Theresultalsoshowsthatthere'sacommonmappingbetweenimagecontenttypeandmeansofexpression.
Sceneriesarenormallyexpressedthroughromantic,solemn,elegantortemperaturecontrastingstyles,portraitsbynostalgicandblack-whitewaysandartisticpicturesbydecorating,novel,openmagicalones.
CompositionComparetoImageStyle.
Inthecorrespondenceanalysisofthiscom-parison,562effectivesamplesleadedtoasigvalueof0.
005,suggestingasignicantconnectionbetweenimagestyleandcomposition(Fig.
14).
Inthehistoryofhumanaesthetic,nine-squaredandsymmetrichaveoccupiedtheirplaceincomposition.
Famoushistoricalbuildings,fromGothictoChinesestyle,areFig.
10.
MeansofexpressionFig.
11.
LightsandshadeFig.
12.
ImagestyleFig.
13.
Picturetype&Imagestyle188N.
Liangetal.
designedtobestrictlysymmetric.
Cen-tripetal,guide-line,diagonalandframearealsoprevailingmetamorphismofsymmetric.
Theparingofromanticwithsymmetric,traditionalwithvertical,nine-squaredwithtemperaturecontrast,canserveasagoodreferenceforfuturecompositiondesigning.
LightandShadeComparetoImageStyle.
Scarceuniqueandeasydialarethetwomostwelcomestyles.
Thepessimisticnatureofdeepandsolemnandthedirectdenitionofnon-mainstreamcausesthelackofattractiontothemajority(Fig.
15).
Consideringbothdimensions,there'ssignicantrelationbetweenbacklightandsolemn,capturelightandtemperaturecon-trast,regionalexposureandelegant.
Appropriateexposureissuitableformanystyles,includingromantic,humansensations,traditionalnostalgicandeasydial.
CompositionComparetoLightandShade.
SoftlightpicturestypicallyadoptexpressionsofS-shape,triangle,opentypeandcentripetal.
Diagonalandguide-linesaremostlyusedinphotographicsubtraction,whileappropriateexposureinbalance.
Softlightandcontrastofexposurelevelaretotallyoppositeshowninthegure,indicatingthethoroughdifference(Fig.
16).
5ConclusionByextractingfeaturesofthesampleimages,analyzingthecontentsofsemantictags,lookingforcommonfeaturesinpopularimageswhichholdrelativelyhighdegreeofusers'attention,andstudyingthecorrespondingrelationshipbetweeneachlabel;thisessaytendstogureoutwhyusersarepayingmoreattentiontolandscapeimages.
InFig.
14.
Composition&ImagestyleFig.
15.
Lightandshade&ImagestyleFig.
16.
Composition&LightandshadeTheExplorationofUserKnowledgeArchitecture189addition,usersfavorcompositionbalance,nine-squaredformat,withproperexposure,backlightorthewayofcapturinglight.
Besides,usersalsopreferthetraditionalnos-talgia,deepdigniedblackandwhitephotosorportraits;Photostheylikerangefromlyricalromantic,lively,uniquelandscapetothedailytheme;Overandabove,usersarealsointerestedininnovativephotosaswellasopenmagicartphotos.
Thesendingsaresignicantfortheconstructionofphotosharingsiteuserknowledge.
Inthefuture,againstsuchuserswholikesharingphotosonthesephotossharingwebsites,youcanunderstandtherelationshipbetweenthekeythemesoftheirfavoritepictures,thecompositionandexpression,lightandshadow,styleandtone.
Designerscanlearnthepreferencesandneedsofsuchusersthroughrst-handdetailedandreliabledatatoapplytootherdesignsdesignedforthiskindofuser.
Inthisstudy,themethodusedisconstructionofuserknowledgesystembyana-lyzinguserbehavioramongthosewholikesharingpictures.
Thismethodcanalsobeusedinmanyotheraspectsofthebehaviorofkeywords.
Forexample,intheeldofadvertisingcommunication,productpackingdesignandallotherusersknowledgeminingareasrelatedtopictures.
Inthisstudy,theconstructionoftheuserknowledgeminingmethodisdifferentfromthetraditionalmethodofuserexperience.
Asaresult,itcanbeusedinmanyaspectsandeldstoestablishtheuserknowledgesystembasedongeneralcharac-teristicsofdifferentusers'needs,concerns,andthusfacilitatingdesigners'workingprocess.
Whenidentiedcertainfeatureofthekeywordbehavioroftheuser,designercanquicklydrawfromtheuserknowledgebanktondeffectiveandusableresearchdataforreferencetoaidtheirdesigndecisions.
MiningandConstructionofsuchauser'sknowledgesystemcanbetime-consumingintheearlystage.
However,oncetheuserknowledgebankhasbeensetup,itwillnotonlyfacilitatethedesignertoeffectivelyunderstandtheneedsofusersandhelpdecision-making,butalsomakesiteasierformultipledesignersinonesingledesignprojectstounderstandthecommongoal.
Inthisway,thedesignconsistencyamongseveraldesignerscanbeensuredanditsavesdesignerstimeinreducingcommunicationcostsandintheendlargelyimprovesthecommunicationquality.
Thisstudymainlyintroducestheuserknowledge,imageminingmethod.
Whatremainstobeanalyzedistheconstructionofotherpointsoftheuserknowledge,suchastextandsound.
Itisanareawhichstillworthfurtherstudyingandformsgeneralresearchmethodsandtheories.
Theseaspectscanbeusedassubsequentsupplementaryresearchforuser'sknowledgesystemconstruction.
Awell-establisheduserdatabaseisbuiltonboththetraditionalmethodandtheinnovativenewone.
Gettingtounderstandusers'needfrommulti-dimensionalper-spectiveofbigdatamethodaswellasthetraditionalwayofconductinginterview,surveyandfocusgroupseemstobethenewtrend.
However,thisessaydeemsthatthenewmethodofconstructionisfundamentaltothistrendwhilecombinedwiththetraditionalmethodwillmakeitbetter.
190N.
Liangetal.
References1.
McDonald,J.
E.
,Schvaneveldt,R.
W.
:Theapplicationofuserknowledgetointerfacedesign.
In:CognitiveScienceanditsApplicationsforHuman-ComputerInteraction,pp.
289–338(1988)2.
Blandford,A.
,Young,R.
M.
:Specifyinguserknowledgeforthedesignofinteractivesystems.
Softw.
Eng.
J.
11(6),323–333(1996)3.
DeRosis,F.
,Pizzutilo,S.
,Russo,A.
,etal.
:Modelingtheuserknowledgebybeliefnetworks.
UserModel.
User-Adap.
Inter.
2(4),367–388(1992)4.
Tesch,D.
,Sobol,M.
G.
,Klein,G.
,etal.
:Useranddevelopercommonknowledge:Effectonthesuccessofinformationsystemdevelopmentprojects.
Int.
J.
ProjectManage.
27(7),657–664(2009)5.
Bevan,N.
:Whatisthedifferencebetweenthepurposeofusabilityanduserexperienceevaluationmethods.
In:ProceedingsoftheWorkshopUXEM,9,pp.
1–4(2009)6.
Vermeeren,A.
P.
O.
S.
,Law,E.
L.
C.
,Roto,V.
,etal.
:Userexperienceevaluationmethods:currentstateanddevelopmentneeds.
In:Proceedingsofthe6thNordicConferenceonHuman-ComputerInteraction:ExtendingBoundaries,pp.
521–530.
ACM(2010)7.
Law,E.
L.
C.
,Roto,V.
,Hassenzahl,M.
,etal.
:Understanding,scopinganddeninguserexperience:asurveyapproach.
In:ProceedingsoftheSIGCHIConferenceonHumanFactorsinComputingSystems,pp.
719–728.
ACM(2009)8.
Hassenzahl,M.
,Tractinsky,N.
:Userexperience-aresearchagenda[J].
Behav.
Inf.
Technol.
25(2),91–97(2006)9.
Vnnen-Vainio-Mattila,K.
,Roto,V.
,Hassenzahl,M.
:Towardspracticaluserexperienceevaluationmethods.
In:Law,E.
L.
-C.
,Bevan,N.
,Christou,G.
,Springett,M.
,Lárusdóttir,M.
(eds.
)MeaningfulMeasures:ValidUsefulUserExperienceMeasurement,pp.
19–22(2008)10.
Obrist,M.
,Roto,V.
,Vnnen-Vainio-Mattila,K.
:Userexperienceevaluation:doyouknowwhichmethodtouseIn:CHI2009ExtendedAbstractsonHumanFactorsinComputingSystems,pp.
2763–2766.
ACM(2009)11.
Maguire,M.
:Methodstosupporthuman-centreddesign.
Int.
J.
Hum.
Comput.
Stud.
55(4),587–634(2001)12.
Fan,W.
,Bifet,A.
:Miningbigdata:currentstatus,andforecasttothefuture.
ACMsIGKDDExplor.
Newsl.
14(2),1–5(2013)13.
Fisher,D.
,DeLine,R.
,Czerwinski,M.
,etal.
:Interactionswithbigdataanalytics.
Interactions19(3),50–59(2012)14.
Sarmento,L.
,Carvalho,P.
,Silva,M.
J.
,etal.
:Automaticcreationofareferencecorpusforpoliticalopinionmininginuser-generatedcontent.
In:Proceedingsofthe1stInternationalCIKMWorkshoponTopic-SentimentAnalysisforMassOpinion,pp.
29–36.
ACM(2009)15.
Graham,J.
:Flickrofideaonagamingprojectledtophotowebsite.
USAToday,27(2006)16.
Miller,A.
D.
,Edwards,W.
K.
:Giveandtake:astudyofconsumerphoto-sharingcultureandpractice.
In:ProceedingsoftheSIGCHIConferenceonHumanFactorsinComputingSystems,pp.
347–356.
ACM(2007)17.
Liu,S.
B.
,Palen,L.
,Sutton,J.
,etal.
:Insearchofthebiggerpicture:Theemergentroleofon-linephotosharingintimesofdisaster.
In:ProceedingsoftheInformationSystemsforCrisisResponseandManagementConference(ISCRAM)(2008)18.
Sigurbjrnsson,B.
,VanZwol,R.
:Flickrtagrecommendationbasedoncollectiveknowledge.
In:Proceedingsofthe17thInternationalConferenceonWorldWideWeb,pp.
327–336.
ACM(2008)TheExplorationofUserKnowledgeArchitecture19119.
Mislove,A.
,Koppula,H.
S.
,Gummadi,K.
P.
,etal.
:Growthoftheickrsocialnetwork.
In:ProceedingsoftheFirstWorkshoponOnlineSocialNetworks,pp.
25–30.
ACM(2008)20.
Kennedy,L.
,Naaman,M.
,Ahern,S.
,etal.
:Howickrhelpsusmakesenseoftheworld:contextandcontentincommunity-contributedmediacollections.
In:Proceedingsofthe15thInternationalConferenceonMultimedia,pp.
631–640.
ACM(2007)21.
Yongchang,J.
:Knowledgearchitecturebasedonuserexperience:areviewofthebasicprinciplesforknowledgearchitectureinweb2.
0.
J.
ChinaSoc.
Sci.
Techn.
Inf.
5,018(2010)22.
McGinn,J.
,Kotamraju,N.
:Datadrivenpersonadevelopment.
In:ProceedingsoftheSIGCHIConferenceonHumanFactorsinComputingSystems,pp.
1521–1524.
ACM(2008)192N.
Liangetal.