Research-proposal

ResearchProposalApplicant’sname:HuangTianqiTitlePublicOpinionAnalysisKnowledgeEnhancedFinancialDecisionSupportinBigDataEnvironment1.AbstractThebigdataerahasbroughtgreatchallengestopublicopinionanalysisanditsapplicationtofinancialdecisionsupportsystems(FDSS).UndertheWeb2.0environment,theoriginalpublicopinionmethodsanditsdecisionsupportmodulebecameinsufficientfortheapplicationrequirements.Newpublicopinionanalysismethodsneedstobeabletoanalyzenotonlyallcriticalfinancialmarketinformationbutalsomassiveuser-generatedcontent,includingblogs,tweetsandposts.Moreimportantly,FDSSsneedtoextenditsknowledgebasetonewpublicopinionanalysissoastoprovidemorevaluableevidencefordecisionmaking.Theaimsofthisstudyistotheoreticallyinvestigatethefeasibilityofadoptingpublicopinionknowledge,providedbypublicopinionanalysis,inthefinancialdecisionsupportdomain,andtoproposeageneraldesignofsuchknowledgeenhancedFDSS.2.Introduction2.1.ResearchquestionInbigdataera,peoplearefacingmuchmorechallengingandcomplicateddecisionmakingproblemsthanbefore.Theseproblemsinvolveinformationfromvariousaspectsoffinancialmarket.Tohelpfinancialprofessionalsmakebetterdecisions,effectivefinancialdecisionsupportsystemsalwaysincorporatetheanalysisofmanykindsofmarketinformation,frommarketactivityandmovementdata,companyinternalmanagerialinformationtolatestfinancialnewsforaparticularmarketorcompany.Thereweremanyresearchesonadoptingvariousnaturallanguageprocessingtechniquestotheexaminationonmarketinformationsuchasfinancialnews(SchumakerandChen,2009),annualreports(Li,2006)andonlinediscussionboard(DasandChen,2007),inuseofpredictingmarketmovementorstockpricemovement.Oneofthemostpopulartopicistheapplicationoftextminingtechniquesontextualfinancialinformationtopredicttheperformanceofindividualstockorthemovementofthemarket.Atamacroeconomylevel,Peramunetillekeetal.usednewsheadlinestopredictcurrencyexchangerates(PeramunetillekeandWong,2002).Onacompanylevel,Smailovicetal.examinedtweetstopredictstockpricechanges(Smailovicetal.,2012).Therewerealsomanyspecificapplicationresearchofusingtestminingtechniqueslikesentimentanalysistoexaminefinancialinformation.Ruiz-Martinezetal.proposedasemantic-basedsentimentanalysisalgorithmtoconductsemanticsentimentanalysisonfinancialnews,byassigningdifferentdegreesofpositivityornegativitytoannotatedterms,tocalculatethepolarityofthenews(Ruiz-Martinez,Valencia-GarcíaandGarcia-Sanchez,2012).Butinweb2.0environment,thesepartialapplicationofnaturallanguageprocessingtechniquesononeorfewsourcesoffinancialinformationisnolongerefficientandaccurateenoughforfinancialdecisionmaking.AnimportantcharacteristicofWeb2.0isthesocialweb.Usersinteractwithotherusersorapplicationstosharetheirperspectives,opinionsandthoughts.Theyarenolongerauseroftheapplicationorwebsitebutalsoaparticipantofthesocialweb.Thegrowthofuser-generatedcontentisexponential.Web2.0contentisnolongerdiscretebutwithstronginteractiverelationship.Althoughmanydifferentapplicationsofnaturallanguageprocessinginfinancialinformationanalysissystemsyieldpositiveresults,theymainlyrelyonthetextminingtechniquestoanalyzefinancialinformationlinguistically,calculatepolarityandsentimentscores.ItslimitationbecomemoreobviousinanalyzinghugeamountofinterconnectedWeb2.0content.OneofthemostcomprehensivemethodstoexaminefinancialrelatedinformationunderWeb2.0environmentisconductingpublicopinionanalysiswithvariousinformationsource.Bigdatapublicopinionanalysisisintheforefrontofbigdataandpublicopinionanalysisresearch.Itusetestminingandbigdataprocessrelatedtechniquestoextractusefulknowledgefrommassdatainsupportofdecisionmaking.Itiscriticalforfinancialprofessionsorseniormanagementofcompanytogathercomprehensivepublicopinionknowledgeonrelatedsubjectsaswellassummarizationandinsightsonfinancialmarketinformation.Nowadays,decisionsupportmodulesarealwaysembeddedinpublicopinionanalysissystems,togivedecisionsupportevidencethroughvisualpresentation.Ifpublicopinionanalysissystemcanconstructanexpertknowledgebasewhichcanbefurtherincorporatedinfinancialdecisionsupportsystemtogiveoutintellectualstrategysuggestion,itwouldimprovetheefficiencyandaccuracyoffinancialdecisionmakingprocess.Yettherewasnosystematicresearchstudyonthisarea.Drivenbythemotivationtomakeupthisresearchgap,myproposedresearchquestionslistedasfollows:1.Whetheritisfeasibleforbigdatapublicopinionanalysissystemtoconstructpublicopinionknowledgebase?2.Whetheritisfeasibletoincorporatethepublicopinionknowledgeinthemarketinformationanalysisprocess?3.WhetherthepublicopinionknowledgeenhancedFDSSscanprovidemoreusefulandaccurateevidencetosupportfinancialdecisionmaking?4.Iftheanswersofthefirstthreequestionareyes,thenwhatisthegeneraldesignforthepublicopinionknowledgeenhancedfinancialdecisionsupportsystem?Referringtothepreviousstudiesonbigdatapublicopinionanalysisanddecisionsupportsystem(Chengetal.,2010;陈and曹,2013;Caoetal.,2014),Ibelieveitisfeasibletoconstructexpertpublicopinionknowledgebaseandapplyittofinancialdecisionsupportsystemwithahierarchicalmodel.Asadesignscienceresearch,thisstudywillaimtopropo