Understanding Readers Conducting Sentiment Analysis

UnderstandingReaders:ConductingSentimentAnalysisofInstagramCaptionsMingZhanÅboAkademiTurku,Finland+358442313211mzhan@abo.fiRuiboTuKTHStockholm,Swedenruibo@kth.seQinYUEricssonABStockholm,Swedenvincent.yuqin@gmail.comABSTRACTTheadventofmediatransitionhighlightstheimportanceofuser-generatedcontentonsocialmedia.Amongstthemethodsofanalysisofuser-generatedcontent,sentimentanalysisiswidelyused.Nevertheless,fewstudiesusesentimentanalysistoinvestigateuser-generatedcontentonInstagraminthecontextofpubliclibraries.Therefore,thisstudyaimstofillthisresearchgapbyconductingsentimentanalysisoftwomillioncaptionsonInstagram.Supervisedmachinelearningalgorithmswereemployedtocreatetheclassifier.Threeopinionpolaritiesandsixemotionswereultimatelyidentifiedviathesecaptions.Thesepolaritiesprovidenewinsightsforunderstandingreaders,thushelpinglibrariestodeliverbetterservices.CCSConcepts•InformationSystems➝➝Informationretrieval➝➝Retrievaltasksandgoals➝SentimentanalysisKeywordsCaptions;Instagram;SentimentAnalysis;Readers.1.INTRODUCTIONAccordingtoJenkinsandDeuze[18],weareexperiencingaperiodofprofoundmediatransitioninwhichnewmediaarereshapingthemechanismsofcommunication.Communicationbetweenorganisationsandindividualsisintertwined;therefore,discussingonewithoutconsideringtheotherisfutile.Forinstance,consumerssharetheirownstoriesonlinewithotherconsumersandorganisations,whichisconsidereduser-generatedcontentthat,“comesfrompeoplewhovoluntarilycontributedata,informationormediathatthenappearsbeforeothersinausefu