National Kaohsiung First University of Science and Technology
FirstTech Institutional RepositoryNot a member yet
18584 research outputs found
Sort by
Use of Phased Array Ultrasonic Testing and Its Application in Detecting Critical Crack-like Flaws
No full text[[abstract]]設備會因為內容物及操作條件於設備本體造成缺陷。據文獻指出缺陷類型多以減薄及裂痕兩種為主，其中又以裂痕破壞較難預防，其具有幾何特殊、裂痕尖端高應力等特性，對設備強度的破壞程度不得而知。非破壞檢測用來檢測設備上的缺陷狀況，其特性在於不破壞設備本體，不會降低設備強度。而在非破壞檢測中，又以超音波檢測最為廣泛使用，其原因在於能夠立即研判且精確性高。本研究使用相位陣列式超音波(PAUT)做為檢測儀器。其相較其他超音波檢測方式更為精確。其以不同的檢測平面(scan)及檢測角度，定位範圍內的缺陷，因此相對傳統UT檢測能夠有較佳的檢測再現性。此外，相位陣列式超音波相對傳統超音波在檢測儀器上，因功能多且複雜，使得儀器的操作具有一定的門檻，操作不易。然而，非破壞檢查方式雖能檢測出設備上缺陷的位置及尺寸，卻無法直接提供缺陷對設備本體的影響。適用性評估(FFS)，能夠定量缺陷的設備堪用性，對於裂痕評估在Level1是以臨界裂痕長度作為判定；Level2則是計算破壞力學推透過失效分析圖(FAD)判定是否通過；Level3則是透過理論力學方式定義裂痕的應力強度因子。對於Level1臨界裂痕長度能給予非破壞檢測人員較為簡易使用的方法，其原因在於透過量測裂痕的大小在與臨界裂痕相比，便能得知是否通過。臨界裂痕長度相對其他方法，具有簡易及直接性，因此若要即時性判定設備堪用性，為最佳的方法。該長度的意義為會導致裂痕自我成長的關鍵參數，因此為了避免裂痕成長到臨界裂痕長度導致設備強度發生異常，必須於檢測出裂痕時，對其進行臨界裂痕的評估，了解設備是否維修或換新，以降低工安災害的發生。[[abstract]]Theequipmentcanbecausethecontentsandtheoperatingconditioncreatestheflawintheequipmentmainbody.Accordingtotheliteraturepointedouttheflawtypemanyattenuatesandthefissuretwokindsprimarily,inwhichdestroysbythefissuredifficultlytoprevent,ithasthegeometryspecially,characteristicsandsoonfissureacmehighstress,isunknowntotheequipmentintensitydestructiveness.Thenon-destructionexaminationusesforonthecheck-outfacilitytheflawcondition,itscharacteristicliesindoesnotdestroytheequipmentmainbody,cannotreducetheequipmentintensity.Butinnon-destructionexamination,alsobythesupersonicwaveexaminationmostwidespreaduse,itsreasonliesincanevaluatealsotheaccuracyimmediatelyishigh.ThisresearchusePhasedArrayUltrasonicTesting(PAUT)doesfortheinstrumentation.Itcomparesothersupersonicwaveexaminationwaytobemoreprecise.Itbydifferentexaminationplane(scan)andexaminationangle,inlocalizationscopeflaw,thereforetherelativetraditionalUTexaminationcanhavethegoodexaminationreproducibility.Inaddition,PhasedArrayUltrasonicTechnologyrelativetraditionsupersonicwaveoninstrumentation,becausethefunctionaremanyalsoiscomplex,enablestheinstrumenttheoperationtohavecertainthreshold,operatesnoteasily.Althoughhowever,thenon-destructioninspectionwaycanexamineontheequipmenttheflawpositionandthesize,isunabletoprovidetheflawdirectlyactuallytotheequipmentmainbodyinfluence.TheFitness-for-service(FFS),canquotaflawequipmentserviceable,regardingthefissureappraise1istakesthedeterminationinLevelbythecriticalfissurelength;TheLevel2piecesarecalculatethedestructionmechanicstopushpenetrationfractureanalysisdiagram(FAD)todeterminewhetherpasses;TheLevel3piecesarepenetratethetheoreticalmechanicswaydefinitionfissurethestressintensityfactor.Cangivethenon-destructionregardingtheLevel1criticalfissurelengthtoexamineapersonnelsimplerusethemethod,itsreasonliesinthepenetrationgaugingfissurethesizeintocomparewiththecriticalfissure,thencanknowwhetherpasses.Thecriticalfissurelengthrelativealternativemeans,havesimpleandthesubstantively,ifthereforeinstantaneitydeterminationequipmentserviceable,forbestmethod.Thislengthsignificanceforcancausethefissureself-growththeessentialparameter,thereforeinordertoavoidthefissuregrowingcausestheequipmentintensitytothecriticalfissurelengthtooccurexceptionally,mustinwhenexaminesthefissure,toitcarriesonthecriticalfissuretheappraisal,understoodwhethertheequipmentdoesserviceorexchanges,reducesthelaborpeacefuldisastertheoccurrence
A Case Study of Implementation Performance of Scaffolding Erection
No full text[[abstract]]施工架為營建工程施工期間不可或缺的假設工程，隨著工安意識的抬頭，法規對於搭架材料與施作的要求，也不斷地提高，以達到工安零災害。相對的，搭設成本亦將受影響，而營造廠商無不謀求成本降低的方案。本研究以發電廠為例，比較外勞與本勞的作業績效，對三種執行方案：「外包本勞施作」、「自辦泰勞施作」、「自辦本勞施作」進行評估，以找出較佳的方案，作為日後工程報價與施工執行之參考。首先使用Thomas分析法進行生產力的比較，發現泰勞的搭架生產力高於本勞，而影響生產力的主要因子為搭架難度及工作環境條件。其次以層級程序分析法進行方案評比，前三大考量因子依序為「搭設生產力」、「材料初置成本」及「趕工配合度」，評估結果為自辦泰勞較佳，其次為外包本勞，而自辦本勞殿後。[[abstract]]Scaffoldingsareanecessarytemporaryfacilityduringaconstructionproject.Withtheriseofsafetyawareness,therequirementsofregulationsonscaffoldingmaterialsanderectionhavebeentightenedtopreventanysafetyhazard.Ontheotherhand,theerectioncostwillincreaseasaresultandcontractorsareforcedtoseekwaystoreducecost.Takingpowerplantconstructionasanexample,thisstudycomparestheoperationperformanceofforeignandlocalworkers.Threeimplementationsolutions“Subcontractingwithlocalworkers”,“OwnworkforcewithThaiworkers”,and“Ownworkforcewithlocalworkers”areevaluatedtofindoutthebestoneasthebasisofcostingandprojectexecutioninthefuture.First,Thomas’sanalysismethodisusedtocomparetheproductivityofThaiandlocalworkers,showingthatThaiworkershaveahigherproductivitythanlocalworkers.Themajorfactorsaffectingproductivityarethe“difficultyofscaffoldingerection”and“conditionofworkingenvironment”.Second,theimplementationsolutionsareevaluatedusingtheAHP(analytichierarchyprocess)method.Thetopthreefactorsinorderofimportanceare“productivityofscaffoldingerection”,“initialcostofmaterials”,and“cooperationforspeedingup”.Theresultsofevaluationshowthatthebestsolutionis“OwnworkforcewithThaiworkers”,followedby“Subcontractingwithlocalworkers”,and“Ownworkforcewithlocalworkers”isthelast
The Structural Analysis for Clamping Mechanism of Injection Molding Machine
No full text[[abstract]]本研究係針對射出成形機夾模機構中固定模壁與活動模壁之受力變形狀況，利用Pro/ENGINEER軟體之Pro/Mechanica模組進行分析，研究中先透過應變規量測系統以獲得實際應變數值，再予以修正模擬之模壁邊界條件，提升模擬數據之準確度，以期在新設計的條件精確預測出模壁實際受力變形程度。研究結果顯示，固定模壁與活動模壁結構分析與實際量測值兩者差距均低於機台容許之機械加工公差0.01mm，驗證了本研究所建立之射出成形機夾模機構模壁FEM分析，已達可接受之精準度。透過模擬分析可以了解系統元件之受力變形情況，使設計過程中達到減少實際測試次數及提高產品品質，降低成本及縮短產品開發時程目標。本研究並同時進行模壁結構分析，優化結果顯示：活動模壁與原設計比較，剛性增6%，重量減少23%，而固定模壁與原設計比較，剛性增強7%，重量減少1%，結構分析具相當成效，可據以提升模壁結構剛性與輕量化，達到提高安全性及降低成本的設計目標。關鍵字:應變規量測、結構分析、有限元素方法、射出成形機。[[abstract]]Inthisstudy,stressanalysisoffixedandmovablediemoldwallsintheclampingunitofaninjectionmoldingmachinewasperformedusingthecommercialsoftwarePro/ENGINEER(Pro/Mechanicamodule).Thesurfacestrainswerefirstobtainedexperimentally,usingthestraingaugemeasurementsystem.Thesevaluesweresubsequentlyusedforrefiningtheboundaryconditionsofthenumericalmodeltoimprovethesimulationaccuracy.Suchanapproachisexpectedtoyieldaccuratepredictionofthemoldwalldeformationinanactualsystemwithnewdesignparameters.Analysesdemonstratethatforbothfixedandmovablemoldwalls,theerrorbetweenthesimulationresultsandexperimentaldataremainswithin0.01mm,i.e.,withinthemachiningtolerance.Thissuggeststhattheaccuracyofthecurrentanalysisusingfiniteelementmethod,developedformodelingthemoldwallintheclampingunitoftheinjectionmoldingmachine,iswithinanacceptablelevel.Thesimulationanalysisallowsforbetterunderstandingofdeformationofthesystemcomponents.Thisreducesthenumberofphysicaltestsinthedesignprocessandimprovesthequalityoffinalproductsthateventuallyaidinreducingtheproductioncostandtime.Thestructuralanalysisofthemoldwallwasalsoconductedinthisstudy.Itisobservedthatusingamovablemoldwallleadstoa6%enhancementinstiffnessanda23%reductioninweight,whileusingafixedmoldwallleadsto7%enhancementinstiffnessand1%reductioninweight.Theseresultsindicatethatthenewdesigniseffectiveinbuildingastrongerandlightermoldstructurethatassistinachievingthedesigntargetsofsafetyimprovementandcostreduction.Keywords:Straingaugemeasurement,Structuralanalysis,Finiteelementmethod(FEM),Injectionmoldingmachin
A Study on Improvement of SiO2 Process of Color Filter Physical Vapor Deposition for Thin Film Transistor Liquid Crystal Display
No full text[[abstract]]現今為薄膜電晶體液晶顯示器(ThinFilmTransistorLiquidCrystalDisplay，簡稱TFT-LCD)競爭激烈的時代，尤其在大尺寸TV電視朝向高解析、高色域、低價競爭的環境，因此須朝向產能與良率提升及降低生產成本之目標邁進。本研究目的為薄膜電晶體液晶顯示器(ThinFilmTransistorLiquidCrystalDisplay，簡稱TFT-LCD)彩色濾光片物理氣相沉積SiO2製程改善之研究，將應用六標準差之DMAIC(Define-Measure-Analyze-Improve-Control)改善步驟至專案中。且將實驗所得的實驗數據，透過統計軟體Minitab分析，並將重要關鍵因子提出為田口方法之實驗參數，以SiO2製程膜厚之望目特性為目標，且透過直交表和信號雜訊比(SignaltoNoiseRation，S/N比)求得最佳之參數設計組合，再根據信號雜訊比(SignaltoNoiseRation，S/N比)之結果，用變異數分析(AnalysisofVariance，ANOVA)探討各個控制因素，歸納整理出影響SiO2製程膜厚之關鍵因子，並求得最佳製程參數組合來驗證改善SiO2製程良率。本研究最後運用六標準差的製程能力分析(ProcessCapabilityAnalysis)來驗證SiO2膜厚，實驗結果顯示由0.75提升為1.67，證實此分析方法可提升SiO2製程能力，並對於彩色濾光片薄膜製程的生產良率有很大之助益。[[abstract]]TodayistheeraofintensecompetitionforThinFilmTransistorLiquidCrystalDisplay(TFT-LCD),especiallyinlarge-sizeTVtelevisionstowardshigh-resolution,high-colorgamut,low-costenvironments,Improvetheyieldandreducethecostofproductiongoals.ThepurposeofthisstudyistoimprovetheresearchofThinFilmTransistorLiquidCrystalDisplay(TFT-LCD)colorfilterphysicalvapordepositionSiO2Process,andthesixstandarddeviationsofDMAIC(Define-Measure-Analyze-Improve-Control)Improvethestepstotheproject.TheexperimentaldataobtainedfromtheexperimentwereanalyzedbyMinitab,andtheimportantkeyfactorswereproposedastheexperimentalparametersofTaguchimethod.SiO2processfilmthicknessofthepurposeoflookingforthetarget,throughtheorthogonalarrayandSignaltoNoiseRatio(S/Nratio)toobtainthebestcombinationofparameterdesign,andthenbasedontheresultsofSignaltoNoiseRation(S/Nratio),thevariouscontrolfactorsarediscussedbyanalyzingtheVariance(ANOVA),andthekeyfactorsaffectingthefilmthicknessofSiO2processaresummarized.ThebestcombinationofprocessparameterstoverifytheimprovementoftheyieldofSiO2process.?Inthisstudy,sixstandarddeviationsinprocesscapabilityanalysistoverifythethicknessofSiO2.Theexperimentalresultsshowedthatwasraisedfrom0.75to1.67,whichconfirmedthatthismethodcouldimprovethecapabilityofSiO2processandthecolorfilterfilmproductionyieldoftheprocessisveryhelpful
A Study on Customer Loyalty Behavior for Family Life Cycle in Insurance Company: The Moderating Effect of Sociability and Sales Capability
No full text[[abstract]]隨著消費者意識日益高漲的市場環境，為了避免服務品質與顧客期望兩者間產生差距，確立目標客群並創造顧客經驗成為企業脫穎而出的致勝關鍵，尤其是屬於高信譽服務的保險產業受到強烈地衝擊，更因為老齡化社會易動搖家庭結構，故面臨顧客的口碑或再購之忠誠行為降低的風險。因此，本研究探討家庭生命週期之六階段的顧客群，對保險公司的認知感受反應所刺激的顧客忠誠行為發生，並以社交力與業務能力作為調節效果。本研究係以二十歲以上具有親自購買保險經驗的顧客作為研究對象，後續再依據家庭階段劃分顧客群，主要採用問卷調查法，總共回收940份問卷，完成877份有效問卷，再運用AMOS20.0統計軟體進行結構模式分析，其研究結果顯示，整體而言，即包括家庭生命週期之各階段的顧客群，在服務品質對顧客經驗、顧客經驗對口碑意圖，以及顧客經驗對再購的三個直接效果皆獲得正向的顯著影響，而顧客經驗同時亦具有中介效果的存在，口碑意圖以及再購則是會受到社交力與業務能力的調節影響，隨著不同客群而獲得顯著或是不顯著的差異性結果。本研究的結果揭發顧客經驗的潛在價值，也能夠釐清各階段的客群會因為家庭背景而有不統一的忠誠行為，並透過顧客與業者的視角，進一步瞭解產生行為的過程。最後，本研究依據研究發現，討論行銷管理意涵，針對產業與公司提出以顧客導向為主的具體性建議，使得有助於改善保險產業的市場風險，以提升保險公司的顧客保留力與影響潛在的新顧客，為求研究的真實性，更是提出未來研究建議。[[abstract]]Inordertonottoincreasethegapbetweenservicequalityandcustomerexpectationsduetotheincreasingcustomerawarenessinthismarketchangingenvironment,itiscriticalforenterprisestoestablishtheirtargetcustomersandtoenhancecustomerexperience,especiallyfortheinsurancecompanieswhosehighservicecredibilityhasgreatimpactontheirbusinesses.Inaddition,familystructurenowadayshasshiftedintheagingsocietyandthereisdecliningcustomerloyalty,intermsofword-of-mouthandrepurchase,forinsurancecompanies.Therefore,thisstudyexplorescustomers’familylifecyclereflectstheirloyaltybehaviortowardinsurancepurchase,andtheeffectsofcustomers’sociabilityandsalescapabilitymoderatingtheirattitudesandbehaviors.Thestudyisbasedontheexperienceofcustomerswhohavepurchasedinsurancesexperience.Structureequationmodeling(SEM)withAMOS20.0wasadopted.Thefindingsarethatcustomerswithdifferentfamilystageshavepositiveimpactsofservicequalityoncustomerexperience,ofcustomerexperienceonword-of-mouthintention,andofcustomerexperienceonrepurchase.Differentlevelsofcustomerexperiencealsohavemediatingeffectoncustomerinsurancepurchaseloyalty.Familylifecyclealsohasdifferenteffectsoncustomerword-of-mouthintentionandrepurchasebehavior.Thefindingsofthisstudynotonlyrevealtheessentialroleofcustomerexperienceoncustomerloyaltybehaviorininsurancecompaniesbutalsoclarifythevariousstagesofcustomersintheirfamilylifecycleshowdifferentimpactsontheirloyaltybehavior.Basedonthefindings,thisstudysuggestedinsurancecompaniesandtheirsalespeoplefewpracticalmarketingmanagementandcustomerorientation
An Application of PZB model to the Innovation service of Coast Guard-The Case of Kaohsiung Area
No full text[[abstract]]在現今各公部門逐漸重視服務品質下，服務能量與服務效率都在不斷地成長。然海巡機關在客戶服務的發展相較於其他公部門、民間機構仍屬新興機關，有關「海域執法」、「海洋事務」、「海事服務」三大核心工作的服務品質，其重要性不容忽視。基此，本研究將探討「應用PZB服務品質概念模式對海巡署服務創新之研究」之變化，藉以檢討如何改進之必要模式。本研究以PZB服務品質概念模型作為研究架構，並以南部地區巡防局內部控制與管理做為研究主體，同時採以半結構式訪談，深入瞭解各安檢所成立迄今轉型為全方位、多元化之安檢所模式，來深入說明基層單位如何從執法導向轉型至服務導向，以及依法行政前提下的服務創新成功因素，最後根據研究結果發現，在完整的工作規範下，並展現創新思維推動於服務品質上，均大幅提升服務品質滿意度，在面臨海巡機關組織改造之際，希藉以供服務品質工作參考價值。[[abstract]]Today,moreandmoregovernmentsectorsarefocusonservicequality,sothatservicequalityandefficiencyaregrowinginmanyorganizationconstantly.Andhence,CGAiskindofemergingorganizationindevelopingcustomerservice,whichcomparewithothersgovernmentsornon-government-organization(NPO).MaritimeLawEnforcement,MaritimeServices,andMaritimeAffairsofCGA’sservicequalityisnotonlythemajorcoreworkbutalsosignificantquestionshavetoconsider.BesidesontheresearchisgoingtodiscussthechangeoftheapplicationofPZBmodelservicequalitytoCGA’sserviceinnovationandtrytoevaluatehowtoimprovetheprogress.ThisessayhasappliedthatthePZBmodelasresearchframeworkandthatutilizetoidentifyinternalcontrolandmanagementinbureausouthernpatrol.Otherwise,thesemiinterviewstructurehasbeenadopted.Thisisbecausethatinordertorecognizetransmissionandthathowinspectionofficehasbecomemulti-functionalanddiverse.Moreover,weusethisresearchframeworkandsemistructureinterviewtoexplainsuccessfulfactorsofcreativeservice,whicharehowbaseunittransitsfromlaw-enforcement-orientedtoservice-oriented.Finally,accordingtoanalyticalresults,thesatisfactionofservicequalitywouldberaisedsignificantlyunderperfactruleandcreativesense.Furthermore,Iwishthattheresearchcouldprovidethevalueforreferenceofservicequalityduringthereconstructionoftheorganization
Exploring a Multi-Screen Advertising Effectiveness Based on User’s Perspective
No full text[[abstract]]隨著網際網路、科技的進步與硬體的發展，廣告的推播已經有別於以往，不再受限於紙本的傳播媒介，轉為透過以近幾年發展起來的智慧型裝置結合雲端服務技術來進行廣告的推播，但由於這兩項崛起的產物還未被充分運用，且從產業的調查報告來看也可得知，在多屏的廣告上仍持續成長中，且以行動廣告的成長率最為驚人，故對於一雲多屏廣告的領域加以研究，藉以使用者觀點來探討出何種呈現效果能夠有效吸引到廣告接收者。本研究主要在探討廣告型態、載具大小、廣告效益之間的影響效果，以及廣告表現方式對於廣告效益的干擾效果。而本研究為使更加貼近使用者觀點，故採用實驗法，使受測者配戴腦波儀器並在臉部情緒偵測下觀賞實驗廣告，之後再輔以問卷填答，各別收集到主觀與客觀的數據加以分析。最後資料的分析透過IBMSPSSStatistics22的套裝軟體以敘述性統計、獨立樣本T檢定、單向ANOVA檢定等，分析出彼此的影響與干擾情況，以提出本研究建議給予業者參考，或後進學者進行相關研究時參考。本研究之結果如下：1.廣告型態其廣告效益為：影片類最佳、圖像類其次、文字類最差。2.看板載具廣告效益較手機佳。3.廣告表現方式干擾效果為：產品式表現方式優於情境式表現方式，且看板載具搭配影片類廣告效果最佳，看板載具搭配圖像類廣告效果其次。4.產品式表現方式對廣告記憶干擾效果顯著。[[abstract]]WiththeInternet,technologyadvancesandhardwaredevelopment,advertisinghasbeendifferentfromthepreviouspush,nolongerlimitedtothenewspapersandmagazines.Inrecentyears,throughthesmartdevicewiththecloudservicetechnologytopromoteadvertising,buttheriseofthesetwoproductshavenotbeenfullyutilized.Fromtheindustrysurveyreportcanalsobeseenthatthemulti-screenadscontinuetogrow,andthegrowthrateofthemobileadvertisingismostalarming.SousetheUser’sPerspectiveinthefieldofacloudofmulti-screenadstoexploretheeffectofwhichcaneffectivelyattracttotheadvertisingrecipient.Thisstudymainlydiscussestheeffectofadvertisingtype,vehiclesize,advertisingeffectiveness,andtheinterferenceeffectofadvertisementexpressiononadvertisingeffectiveness.Inordertomakethestudymoreclosetotheuser'spointofview,theexperimentalmethodisusedtomakethesubjectswearthebrainwaveinstrumentandobservetheexperimentaladvertisementsunderthefacialemotiondetection,andthensupplementedbyquestionnairestocollecttheobjectiveandsubjectivedatatobeanalyzed.Finally,theanalysisofthedatathroughtheIBMSPSSStatistics22,suchasnarrativestatistics,independentsampleTtest,one-wayANOVAtesttoanalyzetheimpactofeachotherandinterference,andtoproposethestudyrecommendedtotheindustryreference,orlaterscholarsthestudyreference.Theresultsofthisstudyareasfollows:1.Advertisingtypeofitsadvertisingeffectiveness:Thebestisfilmclass,followedbytheimageclass,thetextclasstheworst.2.Kanbanadvertisingeffectivenessbetterthanmobilephone.3.Theperformanceofadvertisinginterferenceeffectiveness:Product-styleexpressionbetterthanSituational-styleexpression.TheeffectivenessthatKanbanvehiclewithvideoadisthebest,andthatKanbanvehiclewithimageadisthesecond.4.Product-styleperformanceoftheadvertisingmemoryinterferenceeffectivenessissignificant
Twitter Sentiment Analysis by Using Convolutional Neural Networks and Long Short Term Memory Networks
No full text[[abstract]]輿情分析是將社會大眾對事物或觀點，發表自己的意見，蒐集並分析，從而得到有價的資料，進而可以對於決策的走向能更清楚、快速。隨著網路科技發展，取得輿情資料的管道變得更多元化，現今最熱門的不外乎是社群媒體。社群媒體最能表現出民眾最原始的態度，沒有經過任何的修飾，也能最即時得知，而使用人數也一直在增加。因此我們使用twitter做為資料來源。本研究是利用深度學習演算法的卷積神經網路(CNN)和長短期記憶神經網路(LSTM)作為分類器，並將取得的內文轉換成一組向量值，並搭配情緒標籤來進行實驗。本實驗使用SandersAnalyticsCorpus和Sentiment140這兩份資料集作為實驗數據。透過前處理方法將資料集內之hyperlinks、usernames、hashtags、stopwords、emotions、punctuation等多餘的資訊去除掉，以減少影響實驗結果因素。並將兩組資料合併成符合實驗規則的格式。實驗模組使用卷積神經網路(CNN)和長短期記憶神經網路(LSTM)分類其為基本模組，並搭配word2vec、n-gram、CNN+LSTM等進行不同模組的比較。本研究設置三個實驗；實驗一使用5,000筆訓練資料集和450筆測試資料集，實驗結果顯示最高的同時為CNN迭代1次和LSTM迭代20次，取2-gram的準確率為72.14%;實驗二使用1,600,000筆的訓練資料和1449筆測試資料集，結果為LSTM+word2vec進行迭代一次的準確率最高為77.08%。;實驗三則是使用5000筆訓練資料集和359筆測試資料集，用多層感知神經網路(MLP)分類，一層隱藏層(250個神經元)的準確率達到75.21%。[[abstract]]Sentimentanalysisisusedtogetvaluableinformationbasedonthesocialmedia.WiththedevelopmentofInternettechnology,therearemorewaystogetsentimentdatainwhich,themostpopularissocialmedia.Itisabletoshowtheattitudeofthepeople,withit'simmediateandincreasingusers,sohereweusetwitterasoursourceofinformation.Inthisstudy,weusethedeeplearningalgorithm:ConvolutionNeuralNetwork(CNN)andLongShortTermMemoryNeuralNetwork(LSTM)asclassifiers,andconvertthetextintoasetofvectorsandexperimentwithemotion'slabel.WeuseSandersAnalyticsCorpusandSentiment140asdatasetandusepre-processingmethodstoremoveextracontentlikehyperlinks,usernames,hashtags,stopwords,emotionssymbols,punctuationandotherredundantinformationtoreducetheimpactofexperimentalresults.Wemergetheabovetwodatasets.TheexperimentalmodulesuseConvolutionNeuralNetworkandLongShortTermMemoryNeuralNetwork,andcombinedwithword2vec,n-gram,CNN+LSTMtoformdifferentmodules.Inthisstudy,threeexperimentswereimplemented.Experiment1used5000trainingdataand450testingdata.TheresultsshowthattheCNNandLSTMwith2-grammoduleshasthehighestaccuracyof72.14%.Inexperiment2,weused1,600,000trainingdataand1449testingdata.Thehighestaccuracywas77.08%withLSTM+word2vec.Inexperiment3,WeusedMultilayerPerceptron(MLP)forclassify,thehighestaccuracywas75.21%withonehiddenlayerof250neurons
The Impact of Student’s Learning Efficiency and Career Development by Using Library Resources of the National University of Science and Technology
No full text[[abstract]]Carter,AllanM.指出：「圖書館是大學的心臟，在所有非人為的因素中，與高等教育的品質是最為密切相關。」(Allan,1966)，圖書館是大學學生解決課堂上問題、學習專業知能和閱讀書籍的重要地點，但在我國教育部的統計資料中，發現了學生對於大學圖書館的需求日漸減少，於是本研究希望能找到學生使用圖書館是有助於學習成效的，進而增進圖書館的使用率。大學是學生生涯發展中的一個轉捩點，而圖書館的服務除了提供大學學生在校期間的幫助外，亦提供了學生許多生涯發展的書籍及活動，使學生在畢業後的道路上能藉由書籍的知識及活動中所見識的經驗，找到自身畢業後的方向。本研究基於上述所述，希望能探討出學生使用圖書館與畢業後發展的關聯性。本研究先將圖書館提供的服務相關紀錄進行收集，再以文獻佐證挖掘出能表現出學生實質上的學習能力及學生畢業後的實質表現的相關數據，之後本研究將以學生的在學成績做為評估實質上學習能力的依據；接著取得教育部畢業一年後之流向調查問卷的數據中的月薪做為學生畢業後的實質表現，透過將圖書館的數據、學生的在學成績和學生畢業一年後的月薪表現結合整理，來探討學生使用圖書館的習慣差別對學生在學成績和學生畢業一年後的月薪表現之差異及關連性。研究結果顯示：一、圖書館相關服務是影響到學生學習成效的因素，經常使用圖書館資源的同學其在學成績的表現會優於較少使用圖書館資源的同學。二、圖書館相關服務對學生畢業一年後的月薪表現在本次收集的資料中是無法預測的。[[abstract]]Carter,AllanM.said:’’Thelibraryistheheartofauniversity;noothersinglenon-humanfactorisascloselyrelatedtothequalityofgraduateeducation.’’(Allan,1966)Thelibraryisanimportantplaceforcollegestudentstosolveproblemsinclass,learnprofessionalknowledgeandreadbooks.However,inthestatisticsofoureducation,wefindthatstudents'demandforuniversitylibraryisnotasgoodasbefore.Sothisstudyhopestofindstudentstousethelibraryishelpfulinlearning.Universityisaturningpointinthedevelopmentofstudentcareer.Libraryservicesinadditiontoprovidingcollegestudentsduringschoolhelp,butalsoprovidesstudentswithmanycareerdevelopmentbooksandactivities.Thatstudentsontheroadaftergraduationbythebooksknowledgeandtheactivitiesexperience.Findtheirownaftergraduationdirection.Thisstudyisbasedontheabove,hopingtoexploretheuseoflibrariesandgraduatedevelopmentaftertheassociation.Thisstudyfirstcollectstherelevantrecordsoftheservicesprovidedbythelibrary,andthenfindtherelevantdatathatcanshowthestudents'reallearningabilityandthestudents'physicalperformanceaftergraduation.Thisstudywilltakethestudent'sacademicperformanceasthebasisforassessingthereallearningability.ThengettheMinistryofEducationaftergraduationayearlatertothequestionnairedatainthemonthlysalaryasastudentaftergraduationofthephysicalperformance,throughthelibrarydata,studentachievementandstudentsgraduateforoneyearafterthemonthlysalarytocombine.Explorethestudentsusethelibraryhabitsofthedifferencesinstudentachievementandstudentsgraduateoneyearafterthemonthlysalaryofthedifferenceandrelevance.Researchshows:1.Library-relatedservicesarefactorsthataffectstudents'learningoutcomes.Studentswhoregularlyuselibraryresourceswillperformbetterthanthosewhodoesn’tuse.2.Library-relatedservicesforstudentsaftergraduationayearafterthemonthlyperformanceinthiscollectionofinformationisunpredictable
Research on Virtual Community knowledge Sharing of Performance
No full text[[abstract]]網路的發展快速與進步，逐漸帶動虛擬社群的趨勢，而社群的種類也越來越多選擇化，也提供了許多知識分享的機會，在虛擬社群中知識分享是一種傳遞資訊的行為，而這樣的行為在社群中對使用者來說是一種習慣的行為，使用者因興趣相投而組成知識社群來分享知識，在日常生活中充滿各式各樣虛擬社群型式，漸漸取代了用面對面溝通進行分享知識，反而在虛擬社群中分享知識，但是什麼因素讓使用者有意願在虛擬知識社群進行知識分享產生知識績效呢？本研究以知識轉換(knowledgetransfer)與知識利用(knowledgeutilization)作為架構主軸，並透過社會資本理論與任務科技適配度理論為基礎，提出任務平衡度與團隊學習常規等觀點來探討使用者在虛擬社群知識分享績效之研究，以針對參與過虛擬知識社群之使用者作為研究樣本，並以網路問卷調查進行發放並分析，問卷回收樣本數共354份有效問卷，所以該研究發現，(1)知識轉換與知識利用具有正面影響知識績效。(2)任務平衡度會正面影響知識轉換與知識利用。(3)團隊學習常規會正會影響知識轉換與知識利用。過去對於虛擬社群分享知?的研究，大多著重於企業或校園等特定組織內的知?分享，此研究帶來新的驗證。[[abstract]]TheNetworkdevelopmentandprogress,graduallydrivingthetrendofvirtualcommunities,andthetypeofcommunityisalsomoreandmoreselective,alsoprovidesalotofknowledgesharingopportunities,Inthevirtualcommunityknowledgesharingisakindoftransmissionofinformation,Andthisbehaviorinthecommunityfortheuserisahabitofbehavior,Usersbecauseofinterestandcompositionoftheknowledgecommunitytoshareknowledge,Inthedailylifeisfullofmanyvirtualcommunitytype,graduallyreplacedwithfacetofacecommunicationtoshareknowledge,Instead,inthevirtualcommunitytodoknowledgesharing.Butwhatfactorsmakeuserswillingtoshareknowledgetoperformanceinvirtualcommunities?Thisstudy,knowledgetransferandknowledgeutilizationareusedasthemainaxisofthestructure,andbasedonthetheoryofsocialcapitaltheoryandtasktechnologyfit.Andbalancebetweenchallengesandskillsofthetaskandtheteamlearningnormstoexploretheuserinthevirtualcommunityknowledgesharingperformance.Totargetuserswhoparticipatedinthevirtualknowledgecommunityasastudysample,andpublishedandanalyzedbythewebsurvey.Thequestionnairewascollectedandanalyzed,and354validquestionnaireswerecollectedfromthequestionnaire.Thestudyfoundthat:(1)Knowledgetransferandknowledgeutilizationhaveapositiveimpactonknowledgeperformance.(2)Balancebetweenchallengesandskillsofthetaskandtheteamlearningnormswillhaveapositiveimpactonknowledgetransferandknowledgeutilization.Tothevirtualcommunitytoshareknowledgeofthestudy,mostlyfocusingonknowledgesharingwithinspecificorganizationssuchasbusinessesorcampuses,pastacademicrelationshipquicklyapplyonlysustainedbuyingintention,thisstudyalsobringsnewverification