結構化與非結構化資料| Seagate 台灣

裝置或軟體將移往資料湖的原始格式資訊收集後，原始的輸出資料便是非結構化資料。

結構化資料會以數值或文字格式整理妥當，可在預先定義的參數內將資料分門別類、重組分析。

RelatedTopics OptimisedArchitecturesWhitepaper DownloadWhitepaper CloudConfidenceReport CloudConfidenceReport Datasphere2021-CORTX WatchonDemand BigDataAnalyticsSolutions Structuredvs.UnstructuredData Structuredvs.UnstructuredData Inthisarticlewereviewthetwotypesofdataandthedifferentuses.Unstructureddataistherawoutputofdevicesorsoftwarethatcollectinformationwhichismovedintodatalakesinitsoriginalformat.Structureddataisorganisedinnumericalortextformat,andcanbecatalogued,organised,reorganisedandanalysedwithinpre-definedparameters. DefiningStructuredvs.UnstructuredDataMainImage DefiningStructuredvs.UnstructuredDataTherearetwowaysinwhichdataisclassifiedforthepurposesofstorage,analysis,andbusinessdecision-making:structuredandunstructured.Thedifferencebetweenstructuredandunstructureddependsonwhetherornottheinformationisorganisedforthepurposesofdatausageandanalysis.Structureddatatypicallyconsistsofclearlydefinedinformation(likehardtextandnumbers)thatiseasilysearchableandmaintainedinortrackableviaahighlyorganisedtableordatabase.Meanwhile,unstructureddatacomesinavarietyoffileormediaformatsandisn'tintrinsicallyneatlygroupedorclassified.Butthedifferencesbetweenstructuredandunstructureddataextendbeyondhowtheinformationiscollated.Forthepurposesofanalysis,eachrequiresadifferentsetoftechnologytoolsandanalyticalmethodologiesdeployedbydataprofessionalswithvariedknowledgeandskillsets.Organisationstendtoutilisestructureddatamorethantheydounstructured.About43%ofalldatathatorganisationscapturegoesunutilised,representingenormousuntappedvalueinregardtounstructureddata.Butbothdatatypesarevaluableandcanbeexploitedaslongasorganisationsunderstandhowtheydiffer,andthecapabilitiesrequiredtomakeuseofthem.WhatisUnstructuredData?Unstructureddataisinformationinitsrawformat;itoftenlivesinorneartheoriginallocationinwhichitwascollected,orindatalakes—relativelyundifferentiatedpoolsofdata.Becauseitrepresentsalltypesofrawdatathat’scollected,eventhatwhichhasn’tbeencataloguedoranalysed,itrepresentsmassivequantitiesofpotentialvalueandthusrequiresrobustdatacentreandcloudarchitecturesdeployingveryhigh-capacitydatastoragesystems.Thus,unstructureddataishard-driveintensive.Theneedtouncovergreatervaluebyretainingvastquantitiesofunstructureddatainaneconomicalwaymeansthereishigher-than-everdemandformass-capacitystoragesystemscentredaroundharddrives—whichcontinuetoprovidesignificantTCOadvantages,asadvancesinHDDtechnologycontinuetomakeever-highercapacitiespossible.Theneedtoaccessunstructureddatanearitssourceandtomoveit,asneeded,toavarietyofprivateandpublicclouddatacentrestobeusedfordifferentpurposes,isalsodrivingtheshiftfromclosed,proprietary,andsiloedITarchitecturestoopen,composable,hybridarchitectureswheredatamovesfreelyandefficientlyacrossthedistributedenterprise.Unstructuredinformationisalsoreferredtoasqualitativedata,meaningthatitsimplyinformationthatisobservedorrecorded.InternetofThings(IoT)sensorsinafactory,forinstance,mightcollectdataabouttheongoingperformanceofequipment.Theinformationisthensenttoserverstobestoredinanunstructuredformat,suchasaPDFandvideofiles.Otherexamplesofunstructureddataincludesatellitephotos,weatherreports,patients’biosignaldatainahospital,anddigitalcameraimagerythathavenotyetbeentaggedorcataloguedinanorganisedway.Thecommondenominatoristhatdataispassivelygatheredandtransmittedwithoutanypre-definedorganisationalformatting.Whileunstructureddatahastheopportunitytobeextremelyusefulinspottinglargertrendsandconstructingpredictivemodelswhenithasbeenreviewedandunderstoodaspartofamassivedataset,it'sdifficulttoreadilysearchandanalyseforthepurposesofbusinessanalytics.WhatisStructuredData?Structureddataisorganised,quantitativedata—mostcommonlynumericalortext-baseddata—thatexistsinsomekindofstandardformattinginafixedfieldwithinafileorrecord.Informationthatexistsinspreadsheetsorrelationaldatabasesarecommonexamplesofstructureddata.Thisorganisationmakesitsimpletoquerythedatawhenlookingforspecificpiecesofdataorgroupsofinformation.Forexample,agriculturalsensorsonafarmmightcollectrawweatherdatatodeterminewhencropsshouldbewateredandhowmuchwatertheyneed.Inorderforthedatatobestructured,itneedstobecategorisedandformatted.Thistypeofdatainastructuredformatmightlooklikeatablewithcolumnsentitled“timeofday,"“temperature"and“humidity."Thestructurefacilitatessearching,sortingandanalysing.Structuredvs.UnstructuredDataThemaindifferencebetweenstructuredandunstructureddataistheformatting.Unstructureddataisstoredinitsnativeformats,suchasaPDF,videoorsensoroutput.Structureddataispresentedstrictlyinapredefinedformorwithpredefinedsignifiersthatdescribeit,inastandardizedformatsothatitcanbeeasilyplacedintoatable,spreadsheetorrelationaldatabase.Unstructureddataisoftenhousedinwhat'scalledadatalake,whichisessentiallyarepositorythatstoresrawdatainvariousformats.Structureddataresidesindatawarehouses,repositoriesthatonlyacceptdataformattedtopre-definedspecifications.Adatalakeislikeareservoirthatstoresunstructureddataandmayalsostorestructureddata,whileadatawarehousehousesonlyorganisedandformattedstructureddata.Whetherdataisinalakeorawarehouse,theinformationisstoredinsomeformofadatabase.Themaindifferenceisthatstructureddataisstoredinarelationaldatabase,storedinrowsandcolumnsusingorganisedformatslikeStructuredQueryLanguage(SQL),PostgreSQLorMongoDB.Theseformatsmakestructureddatafareasierforusers—ormachines—tosearch,sortandworkwith.Unstructureddata,bycontrast,isstoredinanon-relationaldatabasesuchasNoSQL.Thetwotypesofdataalsodifferinhowtheymaybeanalysed,aswellasthetoolsandpersonnelneededforworkingwithandmanipulatingthem.Unstructureddataistypicallyanalysedbyusingtechniquessuchasdatastackinganddatamining,whichhavebeendevelopedtoworkwithmetadataandcometomoregeneralconclusions.Whenitcomestostructureddata,moremathematicalformsofanalysis—suchasdataclassification,clustering,andregressionanalysis—canbeused.Intermsoftoolsandtechnologies,structureddatafacilitatestheuseofmanagementandanalyticstools.Examplesoftoolsusedtoworkwithstructureddataare:RelationalDatabaseManagementSystems(RDBMS)CustomerRelationshipManagement(CRM)OnlineAnalyticalProcessing(OLAP)OnlineTransactionalProcessing(OLTP)Softwarethatcanworkwithlargedatasetsexistinginmultipleformatsaretypicallyusedformanagingandanalysingunstructureddata.Examplesoftoolsformanagingunstructureddatainclude:NoSQLDatabaseManagementSystems(DBMS)AI-DrivenDataAnalysisToolsDataVisualisationToolsUnstructureddataoftenrequiresmanagementbyawell-trainedexpert,andsoftwaretoolsthathavemoreadvancedAIandpredictivemodellingcapabilities,thanthoseusedforstructureddata.Machinelearningisoneofthestrategiesusedfortheanalysisofunstructureddata.Becausestructureddataisalreadysortedandorganised,thesoftwaretoolsusedtoworkwiththesedatasetsaremoreaccessiblefornon-expertbusinessusers.Forexample,inputs,searches,queries,andmanipulationofdataareoftendoneinaself-servicefashionviaahighlyorganiseduserinterface.UseCasesOneillustrationofhowunstructureddatacanbeemployedisinthewaysensordatafromIoTdevicesmaybeusedforpredictivemodelling.Sensorsonafarm,forexample,areconstantlycollectinganddisseminatingdataabouttheclimate,healthofcrops,andfunctionalityofagriculturalequipment.AItoolscanthenanalysethedataandbuildpredictivemodelsforbettermanagementanddecision-making.AIwithmachinelearningcapabilitiescanlearnfromthesepatternsovertime,producingmoreaccuratemodelswitheachsubsequentanalysis.Unstructureddataintheformofweatherandcropgrowthpatternscanbeanalysedtopredicthowmuchwaterornutrientstheautomatedmachineryshoulddeliverinthefuture.Then,theAIsoftwareconductsanautomatedanalysisandconstructsapredictivemodeltoinformbetterfarmmanagementgoingforward.ThisanalysisisbasedonpatternstheAIrecognisesemergingasitsiftsthroughunstructureddatainmultipleformats,likecropgrowthandsoilnutrientpatternscollectedfromsensors.Structureddataisusedinscenariosthatinvolvequantitativeanalysis.Logisticsandinventorymanagementareareasinwhichstructureddataisusefulinimprovingefficiencyanddecision-making.Warehouseinventoryistypicallyhousedintheformofstructureddatawithcolumnsandrowsinarelationaldatabase.Thisdatacantheninterfacewithinventorymanagementorbusinessanalyticssystemstoinformbothbusinessanddatascienceusers.Users,andtheirsoftwaretools,canplacehardvaluesonmetricsliketheprofitabilityofcertainproductlinesandtheoverheadassociatedwithprocurementandshipping.Companiescanthenmakedecisionsbasedonquantifiableoutputs.Today,thetwotypesofdatahavedifferentuses.Unstructureddataistherawoutputofdevicesorsoftwarethatcollectinformationwhichismovedintodatalakesinitsoriginalformat.Structureddataisorganisedinnumericalortextformat,andcanbecatalogued,organised,reorganisedandanalysedwithinpre-definedparameters.AsAIandMLcontinuetoadvance,newcapabilitiestomine,analyse,learnfromandmakeimmediateuseofunstructureddataarelikelytoemerge. RelatedProducts LyveCloud:StorageasaService LearnMore LyveRack:ConvergedInfrastructure LearnMore BigDataAnalyticsSolutions LearnMore ASeagateReport RethinkData:PutMoreofYourBusinessDatatoWork AboutSeagateOurStoryGlobalCitizenshipTrustCentreQualityAssurance SupportProductSupportSoftwareDownloadsWarranty&ReplacementsContactUs PressCentreNewsMediaKitsBrandPortal Investors PartnersIndirectPartners&ResellersSeagateChampionsSeagateDirect&SuppliersSeagateTechnologyAllianceProgram JobsLifeatSeagateDiversity,EquityandInclusionUniversityProgrammes WheretoBuyProductFinder ©2022 SeagateTechnologyLLCLegal&PrivacyVulnerabilityDisclosure CookiesSettings LaCieMaxtorLyveLabsSeagateGovernmentSolutionsSiteMap ×