雲端運算為近年來新興的熱門話題，強調只要透過網路就可以讓使用者依據實際需求彈性地使用計算資源並按照使用量的多寡計費。許多研究報告認為這種將計算、儲存資源以服務形式提供給使用者的租用模式，將有助於節省資訊投資成本而備受企業主矚目；除了節省成本誘因之外，採用雲端運算服務後可能衍生資訊安全、保密的疑慮，也是大家所關注的焦點，因此使用者如何於採用雲端運算服務的決策過程中取得利弊的平衡點相當重要。 本研究將以企業使用者的觀點，探討影響其使用雲端運算服務的決定因素內涵，範疇侷限於台灣的進出口業者。研究根據過去文獻資料，彙整出五個影響使用者決策因素的構面：知覺效益、安全與信任、創新認同程度、知覺有用性以及知覺易用性，並以問卷調查方式進一步瞭解影響企業使用決策的因素；經由問卷調查的結果發現，上述考量構面足以解釋台灣進出口業者使用雲端服務意願百分之六十的變異，而其中影響業者使用意願程度最高的因素為知覺有用性。 / Cloud computing which emphasizes cloud users can elastically leverage computing resources on demand through the internet has recently become a hot issue. Many studies claimed that this rental model of providing storage resources as a service to users would be good for business to save the cost of investing in information technology which attracts lots of business owners. On the other hand, adoption of cloud computing service may arise several concerns like data security, service reliability and so on. In all, it is really hard to strike a balance during the decision-making process whether should business adopt cloud services. In this research, we explored the determinants of using cloud service form business users’ viewpoints which limited to the import/export industry in Taiwan. Based on past literature, we adopted five dimensions in decision-making: perceived benefits, security and trust, recognition of innovation, perceived usefulness and perceived ease of use. Investigating the findings through questionnaire survey, we found that the above dimensions can explain 60% variation of the causes of adopting cloud services to business users and the most sufficient impact factor is perceived usefulness.

雲端運算是近來全世界最熱門的資訊議題，任何設備只要能夠連上網路，就可以享有各種不同的網路服務，這些服務可能建置在不同的機器設備上，甚至可能在相隔遙遠的雲端裡，因此也稱之為雲端服務，在操作過程中，使用者不需要擔心服務安裝在哪裡，亦不用擔心服務如何達成，可想而知網路服務的背後，隱藏著很複雜的技術與架構。 網路服務的開發方式相當地簡單且快速，為系統帶來更大的便利與彈性，不過在管理的部份，卻變得越來越複雜，例如不易瞭解網路的結構與狀態、潛在的無窮迴圈及冗餘的流程問題，甚至是在資源共享的情況下，因等待或其他因素而導致死結的情況產生。這些問題將使得網路服務組合在執行時期，可能發生無法預期的錯誤。情況嚴重時，系統可能會完全鎖死或停止，對公司將造成重大的財務及商譽上的損失。這些流程的問題，需要在網路服務組合執行前先進行驗證，本研究透過流程轉換成派翠網路的分析，以確保流程設計的正確性與穩定性。 因此，本論文運用網路服務標準，將雲端服務轉換成派翠網路模型，再以派翠網路中的針織法為分析基礎，檢驗模型的狀態，避免死結發生，並提供網路特徵報告書，以降低管理的複雜度，進而提升服務的穩定性。 / Cloud computing is regarded as the most popular recent ICT topic. Any equipment which can be connected to Internet can provide its user have the access to the various services, which may be built on different machines, or even may exist in the distant 'clouds' far away. However, in the operating process, the user has no need to worry where the service locates, needless to say how the service operates, which means that a considerably amount of data and techniques hide behind the Web service. Web service is an artifical piece of art, the user and the manager can concatenate several Web services online into a bigger one according to the conditions they need, which is a considerably simple and fast developing method bringing more convenience and flexibility for the system. Nonetheless, the managerial part also becomes more complex in problems like potential infinite loops and abundant procedures, for instance. What is worse is under the circumstance of resource sharing, the deadlocks happen while pending or other factors occur. These issues will lead to severe errors while the Web service compound operates. If the system completely locks and stops, serious financial damage and loss of commercial reputation will be caused to the company. Hence the design of procedures needs to be validated and transformed as Petri Net analysis before the Web service compound operates in order to assure the accuracy and stability of the procedure design. Therefore, this paper uses Web service standards, turning Web services into Petri Net models with knitting technique as analytical base, to validate the accuracy and stability of the model to avoid deadlock to happen in order to enhance the reliability of the service.

在服務經濟的時代，人們為了提升生活品質越來越重視心靈需求而非過往所重視的物質需求。文化創意產業提供的原創性以及內容扮演越來越重要的角色，文化創意產業近年來在世界各國之經濟地位也漸趨重要；本篇論文提出一個透過網際網路提供台灣國立故宮博物院動態展覽的嶄新影音廣播服務--「故宮影音頻道」。故宮影音頻道意圖透過影音多媒體展示，更全方位且完整地呈現中華文物的精緻特質和內涵，以帶給瀏覽者更清晰具體的文物觀賞體驗，以及激發瀏覽者之創意靈感。 故宮影音頻道為了符合高標準的服務品質以及處理來自世界各地使用者的潛在高峰需求而特別設計負載平衡的機制。本篇論文更進一步地探討故宮影音頻道平台後端作業流程的技術挑戰以及展示提供線上影音廣播服務的具備彈性、靈活、分散特性的雲端平台。首先，我們在雲端平台上建立具有處理大量使用者的負載平衡機制的故宮影音頻道。接著我們讓故宮影音頻道上線進行實驗、收集資料及分析。根據實驗的結果，我們為了維持高標準的服務品質應用排隊理論設計了兩階段的服務系統，最後探討在不同電腦及網路環境下，如何利用該服務系統計算出符合服務品質的設定。 / The iPalace Video Channel is a video-based website and provide people around the world a brand new video-broadcasting service on the Internet which introduces relics in National Palace Museum in Taiwan. The iPalace Video Channel is designed to meet the high standard of service quality which users can watch videos without delay and be capable of coping with potential high peak demands from worldwide. In this paper, first, we establish the iPalace Video Channel and an elastic and reliable mechanism via load balancing that is capable of handling potential huge peak-demand of video services. Second, we conduct an experiment by sampling the video-based service in practice and evaluating the presented approach online. Third, from the finding of evaluation, we establish the two phase service system for delivering video services with high standard of service quality. Finally, we discuss the formal queuing model of video services from which the settings of the queuing system that satisfy acceptable service quality could be calculated under various computer and network environments

為了讓一般的投資大眾能享有智慧型、系統化的策略交易環境，本研究計畫發展一個可供大量使用者共用、並且容易上手的策略交易平台。為了達到這個目的，此平台必須擁有快速且大量的運算能力，雲端運算所提供之大量且可擴充的運算能力，使之成為最適宜的平台。為滿足不同使用者不同的投資偏好，此平台提供多項常用之技術指標與K線型態辨識功能讓使用者利用基因演算法產生符合其偏好的交易策略。在策略產生之後，使用者可以在平台上檢視交易策略在不同商品、不同時間區間上的表現，並從最後的策略回測報告中加以評估，挑選出獲利能力、波動程度與交易頻率都符合需求的交易策略。

隨著雲端運算盛行,企業採用大量虛擬主機來取代實體機器,虛擬主 機有效率的模擬實體機器達到企業減少能源耗損與提高成本效率目 標。 文中提及虛擬主機映像檔目錄系統(VMIC)主要讓使用者能有效 率搜尋期望的檔案並獲得下載的實體位置,故本論文研究重點著重在 改進安全性在原 VMIC 系統,應用 Pakiti 監控系統來掌握更新檔狀況 於實體機器或虛擬機器環境,使資安人員能在短期間內獲得正確資 訊,及時升級更新檔避免攻擊災害發生。 / As cloud computing techniques advance, Virtual Machines (VM) seems to be an appropriate solution than physical machine deployment. Having multiple instances of virtual machines cause more efficient use of computing resources to achieve the aim of energy consumption and cost effectiveness. In this thesis, Virtual Machine Image Catalogue (VMIC) is designed for helping users search and acquire expected virtual machine images promptly. Nevertheless, security of VMIC is also a crucial task to keep systems up-to-date and defends against security attacks. Pakiti is adopted to monitor patch status of physical and virtual machines, and schedules the warning information to remind security staffs to update the patches.

本研究應用雲端分散式的架構來建置與佈署一個處理大量使用者交易需求的高頻交易投資策略服務平台，此平台有以下特色： 1. 系統後端採用雲端SOA架構，將整個龐大的交易系統切割佈署到雲端叢集之上，並提供單一的Façade介面供外部使用者呼叫；系統前端畫面的設計遵循Yahoo UI嚴格的MVC架構規範，並保證前端的View與Model與後端的資料達成同步。 2. 不斷接收來自外部的即時報價訊息，並產生海量的即時市場狀態資訊，包含多種技術分析指標、買賣規則…等，以供高頻交易的策略作為買賣的依據。 3. 利用Java Message Service將大量的即時市場狀態資訊快速、非同步的派送給分佈在雲端叢集各節點的系統模組，並採取Publisher-Subscriber的模式來維持分散後各系統模組之間的鬆散關係。 4. 多樣化的統計演算法模型可供使用者作為產生優良的個人化投資策略之依據。產生的新策略可馬上投入即時的模擬交易環境下監控與評估其策略績效。

現今視頻網站佔電腦使用者使用網路時數極大比例，在世界各地亦有越來越取代電視之勢，然而隨著視頻網站之風行，相關著作權問題亦隨之而生。特別是網路上不少行為常介於合理使用模糊空間，關於法律條文與案例事實之合致，常引發爭論。網路之精神著重分享，然而著作權之保護則屬於國家保護之一種排他權，其與分享之精神，某些時候會有所衝突，如何在網路時代運用著作權保護創作人，卻又不扼殺創意、公益使用，應有探討必要。是以本文將針對視頻網站上開相關問題，作一基本探討，並對外國與臺灣在視頻網站相關立法與案例作整理介紹，特別是美國法立法與相關案例，與我國法制、實務判決比較，並分析檢討後，提出具體看法，或可做為我國未來立法走向以及實務判決參考。 本文架構上，第一章為緒論，內容為研究動機與論文大致架構；第二章則就視頻網站現況與商務模式探討，期能鳥瞰全貌；第三章與第四章分別就視頻網站使用者與業者之責任探討，均以美國法與我國法為主，輔以中國大陸法規做比較與得失分析；第五章則論及新科技演進與著作權法制、法院判決之因應；第六張結論，最末章為參考文獻。

There were many technical events held in 2017, like 2017 COMPUTEX, Google I/O, Microsoft Design Forum…etc., and Artificial intelligence (AI) is one of the most concerned and hottest topics in these events. There is no doubt that in the next few years, AI technology and its further developments will be the major focus in the science and technology industry. As AI technology brings in a lot of new applications and develops commercial values, for the server industry and corresponding supply chain, it will lead to a new wave and benefit the whole ecosystem. Although the growth rate of desktop and laptop computer slows down and smartphone market starts to saturate, the demand for data processing and computing continues to grow. Large datacenter, server hardware and cloud applications supported at the back-end could still keep their momentum of growth. Taiwan-based original design manufacturers (ODMs) maintain a stable growth rate in 90% of the worldwide server manufacturing and many new business models evolved in order to fulfill the demands brought by artificial intelligence (AI). The aim of this study is to analyze the positive and negative impacts of AI technology on server industry, and will focus on four major groups in the industry: datacenter customers, OEM, ODM and component suppliers.

Kyoto University (京都大学) / 0048 / 新制・課程博士 / 博士(工学) / 甲第16839号 / 工博第3560号 / 新制||工||1538(附属図書館) / 29514 / 京都大学大学院工学研究科機械理工学専攻 / (主査)教授 小森 悟, 教授 稲室 隆二, 教授 花崎 秀史 / 学位規則第4条第1項該当

隨著大數據時代來臨，機器學習方法已然成為熱門學習的主題，主要分為監督式學習與非監督式學習，亦即分類與分群。本研究以羅吉斯迴歸配適結果加權距離矩陣，以資料雲幾何樹分群法為主，在含有類別變數的紅酒資料中，透過先分群再分類的方式，判斷是否可以得到更佳的預測結果。並比較監督式學習下各種機器學習方法預測表現，及非監督式學習下後再透過分類器方法的預測表現。在內容的排序上，首先介紹常見的分類與分群演算方法，並分析其優缺點與假設限制，接著將介紹資料雲幾何樹演算法，並詳述執行步驟。最後再引入加權資料雲幾何樹演算法，將權重的觀點應用在資料雲幾何樹演算法中，透過紅酒資料，比較各種分類與分群方法的預測準確率。 / Machine learning has become a popular topic since the coming of big data era. Machine learning algorithms are often categorized as being supervised or unsupervised, namely classification or clustering methods. In this study, first, we introduced the advantages, disadvantages, and limits of traditional classification and clustering algorithms. Next, we introduced DCG-tree and WDCG algorithms. We extended the idea of WDCG to the cases with label size=3. The distance matrix was modified by the fitted results of logistic regression. Lastly, by using a real wine dataset, we then compared the performance of WDCG with the performance of traditional classification methodologies. The study showed that using unsupervised learning algorithm with logistic regression as a classifier performs better than using only the traditional classification methods.