National Kaohsiung First University of Science and Technology
FirstTech Institutional RepositoryNot a member yet
18584 research outputs found
Sort by
Design of a Bicycle Wheel Display System with Imaging Capability
[[abstract]]本論文開發一套裝設在自行車車輪的顯像系統,透過裝設在車輪的LED(Light-EmittingDiode)燈條以及人眼視覺暫留(PersistenceofVision)的原理,令自行車在行駛時車輪看起來像是一個圖案,讓使用者在夜間騎乘自行車時可以大幅提升安全性。本論文以嵌入式系統的概念,結合軟體與硬體來開發自行車車輪顯像系統,利用微控制器單晶片搭配自行設計的控制電路與LED顯像電路,讓車輪在旋轉時可以完整呈現一個圖案,再加入藍牙無線網路技術,令使用者可以透過智慧型手機來更換圖案,我們自行設計的手機應用程式可提供六個圖案讓使用者選擇,還可以利用手機拍攝照片顯示。本系統結合了嵌入式系統、藍牙無線網路技術以及Android手機應用程式(ApplicationProgram,APP),為一套自行車在行駛時車輪可顯示出圖案的顯像系統,實驗結果表示本系統可在自行車時速10-30公里時完整顯示圖案,而時速低於10公里則顯示霹靂燈、閃爍燈以及跑馬燈,讓自行車騎士在夜間騎乘時可以得到完整的保護。[[abstract]]ThisthesispresentsLED(LightEmittingDiode)imagedisplaysystemthatcandisplaypatternsonthespokesofabicycle.TheproposedsystemofLEDlightsthataremountedonthewheelcanprovidesafetybyproducingabeautifulpattern.ThemaincontrolboardandLEDlightingstripsthatdeterminethedisplayedpatternaredevelopedusingembeddedsystemandelectricalcircuitdesigns.Amobileapplicationprogramisdevelopedtocontrolthelightinghardwareremotely;thesystemcommunicateswiththemaincontrolboardviaaBluetoothwirelessnetworkinterface.Thecyclistcanchangethepatternsusingthemobileapplicationprogramorbypushingbuttonsonthemaincontrolboardbeforeriding.Sixpatternsaredesignedfordisplayusingthissystem,andthepatterncanbemadetochangerepeatedlyatpresentintervals.Thesystemalsoprovidestheapproximatedimagedisplayofactualobjectfromtheshootingofmobilephone.Experimentalresultsrevealthattheproposedsystemperformseffectivelyonthewheeluptoamaximumspeedof30km/hr.Whentherotatingspeedofwheeliswithin10km/hr,thesystemdisplaysseveralfixedimagescircularlyforprovidingthesafetyofbiking
Design and Analysis of High Efficiency Conversion Techniques for Ultra-Capacitor Charger Applications
[[abstract]]本論文提出具同步整流輸出之全橋式超級電容充電電路,可實現超級電容瞬間大電流充電之需求,同時保有高轉換效率。轉換器結合倍整流技術,以降低全橋轉換電路之二次側環路電流,提升轉換器效率;同時利用倍整流之特性優化變壓器輔助繞組驅動同步整流開關之訊號。超級電容及其充電電路搭配需供給大電流之負載,可緩衝電源瞬間大功率輸出。實驗電路規格為輸入電壓380V、輸出電壓14.6V、輸出功率1kW實現所提之轉換器架構;最高效率於600W達到93.4%,滿載1kW時效率92.1%;最後以250F之超級電容實測,該轉換器可於一分鐘內將該超級電容充電至14.6V。驗證本論文提出轉換器之可行性。[[abstract]]Inthispaper,asynchronousrectifierfull-bridgedc-dcconverterforchargingultra-capacitorispresented.Thatcanrealizetherequirementofcharginghighcurrentimmediately.Theappliedconvertercombinesthetechnologyofcurrent-doublerrectifiertoreducethesecondarycirculatingcurrent,promotetheconversionefficiencyandamelioratetheself-drivingsignalofsynchronousrectifierswitches.Theappliedconverterwithinput380Vandoutput14.6Visrealized,thehighestefficiencyis93.4%at600W,andtheefficiencyisachieving92.1%atfullload1kW.A250Fultra-capacitorischargedbyappliedconverter,fromzeroto14.6Vin1minute.Theappliedconvertercanbeproveditsfeasibilitybyexperimentalresult
Design and Implementation of Combining Modularized Batteries Charge Equalization and High Step-up Conversion Techniques
[[abstract]]本論文提出應用於高昇壓轉換能源系統之串聯電池快速平衡充電技術。簡化控制之電芯模組整合返馳型與昇降壓型轉換架構,可達成較少元件、快速平衡,並針對高數量單電芯時,電荷不易平衡之缺點改善,且能準確監測各電芯之電壓狀況,以提升電池充電電源管理系統執行之效能。此外,為因應再生能源供電系統,如太陽能、氫能等低輸入電壓,設計一隔離型高昇壓高效率轉換器以匹配串聯高壓電池、380V直流匯流排。串聯電池平衡技術包括電芯模組內平衡與電芯模組間平衡等兩大範疇。電芯模組內平衡器主要由額定功率15W之昇降壓型轉換器構成,電芯模組間平衡器則由額定功率40W之返馳型轉換器構成,並輔以數位運算程式控制系統運行。實測上採用12枚10Ah之磷酸鋰鐵電池串聯進行平衡測試,以驗證本系統之可行性。隔離型高昇壓轉換電路結合昇壓型、返馳型轉換器、倍壓技術,並採用主動箝位機制,能無損地將漏感能量回收至輸出端,進而提高轉換效率,而主開關皆為同訊號驅動,能大幅簡化訊號複雜度,並以實際測試來驗證轉換器效能,測試規格為輸入電壓48V,輸出電壓400V,額定功率800W,本轉換器之最高效率為93.95%,滿載效率為93.6%。[[abstract]]Ahighspeedbatteryequalizationtechniqueappliedinseries-connectionbatteriesforhighstep-upenergyconversionsystemisproposedinthisthesis.Sincethecellmoduleswithsimplecontrolprocesscombinebuck-boostconverterandflybackconverter,fewelementsandfasterbalancespeedcanbeachieved.Theproposedequalizerstructureimprovesthedefectthatthechargebalanceismoredifficultwhentheamountofcellisnumerous.Moreover,theproposedsystemcanmonitorthevoltageofeachcellexactlytoenhancetheperformanceofbatterychargemanagersystem.Furthermore,theisolatedhighstep-upconverterisproposedformatchingtheseries-connectionhighvoltagebatteries,DCBusandtherenewablesourceswithlowvoltage,suchassolarenergyandhydrogenenergy.Theequalizerwithinmoduleiscomposedofbuck-boostconverter.Additionally,theflybackconverterconstitutestheequalizerbetweenmodules.TheperformanceoftheequalizationsystemisverifiedbyusingtwelveLiFePO4batteries.Theisolatedhighstep-upconverterintegratestheboostconverter,flybackconverterandvoltagedoublingcircuit.Theproposedconverterisimplementedwithinputvoltage48V,outputvoltage400Vandpowerrating800W.Further,thepeakefficiencyandfullloadefficiencyoftheproposedconverterareabout93.95%and93.6%,respectively
Design and Implementation of Novel DC-DC Isolated Bidirectional and Unidirectional Converters
[[abstract]]本論文提出兩種新型隔離型直流/直流轉換電路,其一為「新型雙向隔離型電能轉換器」,另一為「新型單向高升壓電能轉換器」。「新型雙向隔離型電能轉換器」結合三繞組耦合電感以及倍壓電路,完成具電氣隔離之高升降壓功能,且無須外加額外主被動電路,即可完成漏電感能量回收。再者,所提之轉換器具備低元件個數,僅以四顆主動開關即可達成雙向高升降壓之動作、全部主動開關柔性切換之功能,故柔性切換技術有助於降低主動開關切換損失並提升電路的轉換效率。「新型單向高升壓電能轉換器」係結合兩組耦合電感與倍壓電路達成電氣隔離、高電壓轉換比之功能,且所提電路同時具備低匝數比、開關柔性切換、漏電感能量回收以及開關低電壓應力之優點。本論文將分析所提之「新型雙向隔離型電能轉換器」以及「新型單向高升壓電能轉換器」的動作原理,其中,包含穩態分析、元件設計與電路軟體(PSIM)模擬,最後,實作電路雛型證明兩個電能轉換電路之正確性。[[abstract]]ThisthesisproposestwoDC-DCconverters.OneisnovelisolatedbidirectionalDC-DCconverter(IBDC),andtheotherisnovelisolatedunidirectionalhighstep-upDC-DCconverter(IUHSDC).TheIBDCincorporatesathree-windingcoupledinductorandavoltagemultipliertofeaturegalvanicisolationandachievehighconversionratio,whichisabletorecycleleakageenergywithouttheuseofactiveorpassivesnubbercircuits.EventhoughtheproposedIBDConlyincludesfouractiveswitches,itcanprocesspowerbi-directionally.ThisswitchnumberistheminimumascomparedwithotherisolatedbidirectionalDC-DCconverters.Withrespecttoconverterefficiency,allswitchescanbeoperatedwithsoftswitching,reducingswitchinglosssignificantly.TheIUHSDCutilizestwocoupledinductorsandtwovoltagemultiplierstofunctiongalvanicisolationandhighconversionratio.TheproposedIUHSDChastheadvantagesoflowturnsratio,lowvoltagestressesonactivecomponents,softswitching,andleakageenergyrecycling.Theoperationprinciple,steady-stateanalysis,designconsiderations,andexperimentalresultsofbothconvertersarepresentedindetail.Prototypesofthetwoconvertersarebuilt,testedandmeasured,whichvalidatetheproposedIBDCandIUHSDC
Establishing a Visualized Finding System - Library and Information Center of National Kaohsiung First University of Science and Technology as an Example
[[abstract]]在這資訊爆發的年代,圖書館亦是廣為人用來獲得資訊的方式,而圖書館於管理與服務的品質亦隨之精進與發展,在手機不斷普及的大環境中,圖書館與行動應用服務亦隨之發展。不過在找書系統中始終僅限於文字的描述,民眾無法更直接的知道書籍位置。本研究便是以資訊視覺化(InformationVisualization)與使用者介面(UserInterface)為基礎並運用MicrosoftVisualStudio2015ApacheCordova為開發工具來建置跨平台應用程式(本研究以Android為開發範例),並搭配後台資料庫做為數據資料的來源,結合WebService技術做跨平台資料存取並進而於手機上體現出視覺化的呈現讓繁雜的文字中賦予新的顯示方式帶給使用者在使用圖書館的找書系統時更具功能性與迅速性,讓較少使用圖書館的使用者也能很快速的找到想要找的書籍位置、入門也更為快速,對於導入此功能的圖書館亦能提升整體服務。本研究以系統設計規劃與展示的研究方法說明如何透過資訊視覺化的方式來提升一般使用者找書的效率,過程以人為中心的方式來做設計與開發,並以問卷調查方式來評估系統接受度與滿意度,用以說明本系統為可提升找書效率之系統。[[abstract]]Despitetheriseininternetusage,manypeoplestillgotopubliclibrariestogatherinformaitonandfindresources.Withrapidinventionanddailyuseofmobilephones,librariesarealsochallengedinsatisfyingtheircustomer'sneedsbyprovidingmoreefficientwaystosearchinformation.However,thebookfindingsystemsnowadaysinmostlibrariesarerestrictedtoliteraldescriptions.Withlimiteddata,usersareoftenfinditdifficultandtime-consumingtogettheprecisebooklocation.Thisstudyadoptstechnologiesincludinginformationvisualization,userinterface,MicrosoftVisualStudio2015ApacheCordova,database,andwebservicetobuildacross-platformapplicationformobilephones(thisstudyisbasedonAndroidsystem).Thisapplicationenablesmobilephoneuserstosearchbooklocationthroughapowerfulgraphicdisplay.Thevisualpresentationmakesiteasierforalluserstodiscoverbooklocationwithinashortperiodoftime.Moreover,withitsuser-friendlydesign,newuserscanalsoutilizethisapplicationintuitively.Inthisstudy,weusetheresearchmethodofsystemdesign,planninganddisplaytoillustratehowtoimprovetheefficiencyoftheusersfindingbooksbyinformationvisualization.Theprocessofuser-centereddesignwaytodothedesignanddevelopment,andassesssystemacceptanceandsatisfactionbyquestionnaires,toillustratethesystemcanimprovetheefficiencyoffindingbooks
Utilizing Data Mining Techniques with R to Discover Business Performances - The Case of the Apple Supply Chain in Taiwan
[[abstract]]蘋果官方網站於台灣時間2017年3月公布2016年全球前200大供應商名單,這些廠商佔全球蘋果公司產品的材料、製造及組裝採購支出至少97%。其中在台供應商計39間公司,惟其中4間非上市或上櫃公司,故本研究僅選擇35間公司作為研究對象,並利用台灣經濟新報資料庫選取其2016年度經營績效指標資料,包含「獲利能力」、「經營能力」、「償債能力」及「現金流量」,以及使用統計軟體R,來執行資料前處理及資料探勘(關聯分析及分群分析),期望探索出有價值的資訊,並達到以下三個目的:一、探討蘋果供應鏈在台廠商,其經營能力指標、獲利能力指標、償債能力指標及現金流量之差異性。二、利用關聯分析技術,了解蘋果供應鏈在台廠商之台灣經濟新報內經營績效指標欄位間之關聯性。三、利用分群分析技術,探討群集特徵並有效地完成資料減量的目的。本研究透過R語言內多種套件執行資料處理及資料探勘,挖掘出隱藏在報表內的資訊,並透過整理及分析結果了解各經營指標之關聯性及群集特徵,以供投資人及相關領域的研究者作為參考。[[abstract]]TheAppleInc.announcedalistofitstop200suppliers,includingcomponentprovidersandothersrepresentingatleast97%ofprocurementexpendituresformaterials,manufacturing,andassemblyofitsproductsworldwidein2016.InthelistofApplesuppliers,39Taiwanesecompanieswereincluded,and35outofthe39companieswerethoselistedcompaniesinTaiwanstockmarkets,includingTaiwanStockExchangeCorporation(TWSE)andTaipeiExchange(TPEx)respectively.Inthisstudy,the35ApplesuppliersoflistedcompaniesinTaiwanstockmarketswereselectedforfurtheranalysisfortheircorrespondingoperationperformancesusinganalyticaltechniquesofassociationanalysisandclusteringwithR.Thevariablesofoperationalperformances,inthisstudy,includeprofitability,operationalcapacity,solvency,statementsofcashflow,andthedataarefromTaiwanEconomicJournalDatabase(TEJ).WithproperanalyticaltechniquesofassociationanalysisandclusteringwithR.thestudyisaimingatexploringanddiscoveringthefollowings:1.TodiscoverthedifferencesofoperationalperformanceswithproperindexfortheAppleSuppliesinTaiwan.2.ToutilizetheanalyticaltechniquesofassociationanalysistoexploretheconnectionsamongthefinancialindicatorsoftheAppleSuppliesinTaiwan.3.ToemployeetheanalyticaltechniquesofclusteranalysistoexplorethesimilarityanddissimilarityofclustersforthoseselectedApplysuppliers.Byfullyutilizingtheinformationdisclosedfromtheannualreportsofthe35ApplesuppliersofTaiwaneselistedcompanies,furtherinsightsoftheinformationconcerningtheoperationalperformancescanbeobtained.Moreover,theinsightsfromtheanalysiscanassistinvestortounderstandthebusinessperformanceofcorrespondingcompaniesintheirportfoliodecisionmaking
The Application of Paragraph Vector Representation with Long Short-Term Memory Neural Network in Sentiment Analysis
[[abstract]]情緒分析是自然語言處理領域中的一個主要挑戰任務。近年來,深度學習(DeepLearning)在自然語言處理的任務中也得到了令人印象深刻的成果。眾多深度學習演算法中的長短期記憶神經網路(LongSort-TermMemory,LSTM)已經在自然語言理解(NaturalLanguageUnderstanding)、機器翻譯(MachineTranslation)、分類與預測(ClassificationandPrediction)問題上獲得了優越的成果。在本篇論文中,本文嘗試了幾種不同輸入模型與分類器來進行情緒分析,包含傳統的詞袋模型(Bag-of-Words)、最新的詞嵌入模型—分佈式詞袋模型段落向量(DistributedBag-of-WordsParagraphVector)。分類器有傳統的邏輯迴歸(LogisticRegression)、與新的深度學習模型—包括長短期記憶神經網路(LSTM)、長短期記憶神經網路加上Dropout正規化(LSTMwithDropout)、雙向長短期記憶神經網路(Bi-DirectionalLSTM)。資料集有兩個,一個是用來做為情緒分析系統的效能基準(Benchmark)的IMDB電影評論資料集,另一個是透過網路爬蟲去Booking.com飯店評論網站爬取得來的真實旅客評論,兩個資料集都是屬於兩類別的分類問題(BinaryClassification)。另外,會同時使用IMDB資料集與Booking.com資料集來驗證各種演算法組合的準確率(Accuracy)。研究結果顯示,在段落向量(ParagraphVector)搭配邏輯迴歸(LogisticRegression)的實驗組合中,以相同的IMDB資料集來驗證詞袋模型(Bag-of-Words)或是段落向量(ParagraphVector)對邏輯回歸(LogisticRegression)的準確率增益的差異大小接近文獻所發表的結果。首先,將資料集更換為自行使用網路爬蟲收集得到的Booking.com資料集,再用完全相同的實驗方式驗證後,發現所得到的結果卻是傳統的詞袋模型(Bag-of-Words)準確率優於段落向量(ParagraphVector)。另一組實驗是使用長短期記憶神經網路(LongSort-TermMemory,LSTM)來對Booking.com資料集進行情緒分類,得到的準確率皆高於邏輯回歸搭配傳統的詞袋模型(Bag-of-Words)或是段落向量(ParagraphVector)。最後,探討長短期記憶神經網路(LSTM)、長短期記憶神經網路加上Dropout正規化(LSTMwithDropout)、雙向長短期記憶神經網路(Bi-DirectionalLSTM),依照三種模型中的平均測試準確率93.89%、最高測試準確率94.61%、最低測試準確率91.61%得出本文的模型組合是長短期記憶神經網路加上Dropout正規化(LSTMwithDropout),這也是最適合用於分析本論文實際爬取的Booking.com的資料集。[[abstract]]Sentimentanalysisisoneofthemainchallengingtaskinthenaturallanguageprocessing.Inrecentyear,deeplearninggetsimpressiveachievementinthenaturallanguageprocessingtask.ManydeeplearningalgorithmincludingLSTMhasgotthebrilliantachievementonnaturallanguagestatisticalmodeling,naturallanguageunderstandingandmachinetranslation.Thisthesistriedseveraldifferentnaturallanguageprocessingalgorithmtoconductsentimentanalysis,includingbagsofwords,distributedbagsofwordsparagraphvector,logisticregression,LSTM,LSTMwithdropoutandbidirectionalLSTM.TheexperimentaldatasetsincludetheIMDBmoviereviewsandtheBooking.comhotelreviewswhichwerecrawledbythisstudy.Bothofthemarethedataforbinaryclassification.AccordingtotheIMDBdataset,theresultoftheresearchillustratesthattheaccuracyoflogisticregressionwiththebagofwordsorparagraphvectorisclosetothatofthepublication.AccordingtotheBooking.comreviewdataset,theaccuracyofLSTMisbetterthanthatofthelogisticregressionwithbagofwordsorparagraphvector.TheaccuraciesofLSTM,LSTMwithdropout,andbidirectionalLSTMareasfollows:averageaccuracyof93.89%,highestaccuracyof94.61%andlowestaccuracy91.61%.ThisstudyconcludesthattheLSTMmodelsaremostsuitableforthesentimentanalysisoftheBooking.comdataset
Design and Implementation of a Distributed In-Place Computing Architecture by Docker Technology
[[abstract]]在此資訊量爆炸的時代,讓大數據資料分析受到了普遍的重視,因此興起了以虛擬機器來建置資料分析平台的趨勢。雖然虛擬機器擁有可快速佈署、充分運用運算資源等優勢,但因為虛擬機器執行速度慢且消耗資源大,因而進一步興起了Docker輕量級虛擬化技術,此技術具備了啟動快、消耗資源低的優勢,並且提供如:整合不同應用程式的Docker-compose、提供叢集功能的Docker-swarm、可快速與分享已建置完成之應用的Docker-hub等完整的工具組合。然而,由於大數據的資料量通常相當龐大,即使有了Docker與相關工具,卻仍不足以應付大量資料傳輸時的效能問題。因此,新創公司MacroData提出了「就地計算」(In-PlaceComputing)的概念:讓大數據資料維持不動,以傳送資料量相對較小的程式碼來取代大數據的巨量資料傳送,降低網路傳輸的瓶頸,又可以兼顧資料在網路上傳送所衍生的資安風險。本研究是使用Docker技術,真正實驗與建置一套就地化資料分析系統,讓中、小企業能以最經濟的方式來實現大數據分析的夢想[[abstract]]Intoday’sinformationexplosiveera,bigdataanalysishasbecominganemergingresearchdirection,whichinspirestheriseofvirtualmachinetechnologyastheinfrastructure.Althoughvirtualmachinegainstheadvantagesofeasydeploymentandmakefulluseofcomputingresources,italsoconsumesalotofresourcesandsuffersfromtheinefficiency.ThatfurthergivesrisetothedevelopmentoftheDockerlightweightvirtualizationtechnology.Thistechnologyisfruitfulbecauseitislowconsumptionofresources,andthecommunityoffersacompletetoolset,suchasDocker-composeforintegratingdifferentapplications,Docker-swarmwithclusteringcapabilities,andDocker-hubforsharingsuccessfully-builtapplications.Besides,duetotheinfeasiblenatureofbigdatatransmissionthroughnetwork,evenwiththeDockerandrelatedtools,itisstillatoughtasktocopewiththetransmissionbottleneck.Fortunately,aninnovativecompany,MacroData,hasproposedtheconceptof“In-PlaceComputing”,whichkeepsthebigdataremainintact,butinsteaddeliversthesoftware(whichissupposedtoberelativelysmall)tothesitecontrollingthebigdata.Hopefully,undersuchcomputingenvironment,thenetworktransmissionbottleneckcanbemitigated,togetherwiththeinformationsecuritycanbeprotected.ThispaperintendtomakeafullutilizationoftheDockertechnologyandrealisticallybuildanexperimenttestbedtohelpTaiwan’ssmallandmediumenterprisetoweavetheirdreamsofbigdataanalysis
A Patent Document Category System by Using Doc2vec
[[abstract]]近年來,隨著資訊化的發展,現今電子檔或數位化類型的文件較過去來的多。專利申請的國家和文件也日漸增多,人們開始減少採用人工的分類的方式進行分類,並提出關於文件自動分類的相關研究方法,以便能幫助管理者或使用者快速分類和找到資料。本研究希望針對由世界知識產權組織(WorldIntellectualPropertyOrganization)所提供的開放資料,來幫助專利權資料文件分類,並提出一種準確率高於過去的新方法。在本研究中,嘗試使用摘要、全文、全文前三百字在去除停用詞後,並使用過去在WIPO專利文件中還沒有人實驗過的Doc2vec進行段落向量的訓練,在調整參數找出模型最佳化的結果後,結果發現利用Doc2vec的DistributedMemory(DM)訓練的文章向量做為特徵值優於DistributedBagofWords(DBOW),並以多層感知器(MLP)等分類器進行訓練並將各個方法的實驗結果作比較,在WIPO-Alpha資料集的Section(第一層)、Class(第二層)、SubClass(第三層)、MainGroup(第四層)的各層中,準確率為73%、85%、95%和94%,在實驗結果中亦表現的比其他方法穩定,各層都有優異的分類結果,並且在研究中,也發現使用全文作為特徵值的分類結果優於摘要和全文前300字,在各個分類器中(SVM、LogisticRegression、MLP、RandomForest),本研究提出的方法中,全文的平均準確率皆高於其他的分類器。[[abstract]]Therearemoredigitalfilesthaneverbeforeinrecentyears.Theproposedmethod(TXT-MLP)introducesanautomaticclassificationmethodtohelpmanagersquicklysortandfindthedata.ThisstudyusesthedatafromWorldIntellectualPropertyOrganization(WIPO)totestandimprovepatentdocumentsautomaticclassification.Ourresearchdatausethreekindsoffeatures:(1)abstract;(2)fulltext;(3)thefulltextofthefirst300words.Doc2vectotrainthemodelandturntheparagraphtovectorforpatentclassification.Afteradjustingparameters,wefoundthebestofparametersintheDoc2vec’smodel.Intheexperimentalresult,theDistributedMemory(DM)methodinDoc2vecisbettertheDistributedBagofWords(DBOW)methodintheWIPO.MultilayerPerceptron(MLP)isusedtobeaclassifier.TheaverageaccuracyofWIPD-AlphadatasetoftheSection(firstlayer),Class(secondlayer),SubClass(thirdlayer),MainGroup(thefourthlayer)is73%、85%、95%和94%.TheMLPmethod’saccuracyarebetterthanothermethods(SVM、LogisticRegression、MLP、RandomForest).Thebestfeatureisfulltextbasedonourexperiment
The Study of Virtual Reality Application in Education Learning - An Example on VR Scientific Experiment
[[abstract]]隨著3D技術、網際網路與多媒體的普及程度,使用者從實體書、電子書到使用網際網路進行線上學習的情形越來越顯著。現在的學習方式除了課堂中聽老師授課講解,還可以到線上平台進行數位學習,甚至使用VR的方式,身歷其境的探索知識,完全跳脫空間、時間的限制。因此,本研究以VR科學實驗為研究範圍,針對體驗價值、使用者態度、使用者滿意度及持續使用意向進行探討,再進一步了解使用者在體驗的過程所受到的美好程度之感受是否會影響其態度、滿意度及持續使用意向,使科技廠商與研發者持續改善並提高使用者態度、滿意度及持續使用意向,以作為VR科學實驗廠商之軟體設計與互動方法之參考。本研究以台灣VR科學實驗之使用者為研究對象,共回收有效問卷103份,輔以深度訪談。採用SmartPLS3.2.6統計軟體進行資料分析與假說檢定,研究結果發現:1.VR廠商或開發者可強化使用者對消費者投資報酬、卓越服務、美感及趣味性之體驗價值,以提升使用者的態度。2.VR廠商或開發者可強化使用者態度,以提升使用者滿意度。3.VR廠商或開發者可強化使用者態度,以提升其持續使用意向。4.VR廠商或開發者可強化使用者滿意度,以提升其持續使用意向。[[abstract]]Withtheuniversalof3Dtechnology,Internetandmultimedia,peoplefrombooks,e-bookstoonlinee-learningismoreandmoresignificant.Thelearningwayintodaycanfromschoolteacher,ande-learningplatform,orusetheVRHead-mountedDisplaytobeimmersedintheexplorationofknowledge,notunderthecontrolofspaceandtime.ThisstudyfocusesontheVRscientificexperiment,asthescopeofresearchtoinvestigateaboutrelatedeffectsonexperientialvalue,user’sattitude,user’ssatisfactionandcontinualuseintention,then,furtherfindouttheinfluencesabouttheniceofexperienceamongusers.ThestudyresultsmayprovideVRtechnologyenterprisestheimprovementofsoftwaredesignandinteractivemethodsofVRscientificexperiment.TheobjectofthisstudywerefocusontheusersofVRscientificexperiment,thereweretotal103validquestionnairescompletedfromVRscientificexperimentusersinTaiwanandusein-depthinterviewtoassist.AdoptingStatisticSystem:SmartPLS3.2.6toanalyzequestionnaireresultsfordescriptivestatistics,hypothesisanalysis.Thestudyresultshow:1.Topromoteuser’sattitudefromusers,thedevelopersofVRsoftwarecanenhanceusers’consumerreturnoninvestment,excellentservice,aestheticsandplayfulness.2.Topromoteuser’ssatisfactionfromusers,thedevelopersofVRsoftwarecanenhanceusers’attitude.3.Topromoteuser’sbehavioralfromusers,thedevelopersofVRsoftwarecanenhanceusers’attitude.4.Topromoteuser’sbehavioralfromusers,thedevelopersofVRsoftwarecanenhanceusers’satisfaction