編輯器針對標的語言提供的編輯輔助，能夠幫助使用者縮短輸入的過程，並且避免輸入的錯誤。然而此種編輯器開發不易，因此大多數特定領域語言(Domain Specific Language)並無提供編輯輔助的專屬編輯器。Eclipse為一套普遍使用的開發環境與工具整合平台，提供了文字編輯器程式框架使建造此種編輯器成為可能，然而大多數的開發者仍然無法透過此複雜的框架開發少數人使用的編輯器。基於此種因素，我們實做EGOE這一套編輯器生成系統，以標的語言的定義為基礎，使用MDA模型驅動的方式在Eclipse平台上自動生成編輯器，使之為標的語言提供結構化編輯與純文字編輯等不同的編輯方式，並且於文字編輯時提供醒目提示、內容輔助與文字喜好設定等編輯相關的輔助功能。 / An editor with special editing aids for a language can increase efficiency and reduce errors while editing sources of the target language. Although of great use, this kind of editors were difficult to create as witnessed by the lack of a language-specific editor (LSE) for most domain specific languages. As the release of Eclipse, however, the situation is changing. Eclipse is an integrated development environment (IDE) firstly published in late 2001 but quickly emerged as a most used IDE. It is indeed also a tool integration platform providing, besides others, a text editor framework, thus making the development of LSEs on Eclipse possible. However, due to the complication of the platform and, especially, the editor framework, it is not easy at all to develop an LSE directly on the framework without any help. We thus in this thesis propose alternatively a generating system of editing environments on the Eclipse platform, which, when given an editing-related specification of the target language, can apply the model-driven approach as suggested by OMG's MDA to generate an editing environment for the target language on Eclipse. The generated editor provides both structured and text editing capabilities. For text editing, we add among its functionalities those such as syntax highlight, content assist and preferences setting for the target language.

投資策略模型有千百種，背後的運算邏輯也大不相同，若是運用不同的演算法進行建模，預測能力必定也不相同。投資人若對標的物僅使用一種策略模型進行預測，其預測力有其上限，考量的因素也無法全面。再進場之後獲利時間點之選擇亦無法精準預測。此時如何能夠同時掌握不同策略模型間的特性和優點，以及找到一套最適當的出場時機規則，為本研究欲解決之問題。 欲解決上述之問題，本研究欲建構出一個即時的投資交易平台，透過多種不同演算法的策略模型，以及同時使用不同時間粒度的歷史資料建出之模型相互參考並輔以權重，使得進場信號的產生不再是僅依靠單一策略，進而提升預測的準確度。此部分將運用Apache Storm的分散式架構來進行實作，從接收市場每秒鐘的即時報價、運算完該秒鐘上千種的市場狀態、再到多個策略間彼此產出信號，僅需要花費毫秒級別的時間即可完成，並轉交給券商下單。而搓合的結果則是透過Kafka的訊息傳遞機制來實現，系統以訂單的狀態來判定最佳的出場時機。

行動裝置在現在這個世代相當普遍，而我們需要一個方法來探索App在背後的行為。 本研究提出了一個非監督式的分群方式，目的是在於探討我們是否能使用App中的原始碼當作以行為分群的依據。 在此研究中，我們應用了迭代分群的方式對Apps做分析，並且觀察分群的結果是否恰當。 而在實驗中，我們由App Store下載了數百個App並加以分析，我們發現我們所提出的方式表現相當良好並且能給出正確的分群結果。 / Smart devices are everywhere nowadays. Mobile application (app) development has become one of the main streams in software industry with more than millions of apps that have been developed and published to billions of users. It is essential to have a systematic way to analyze apps, preferably on their executable that are the only public available sources of apps in most cases. In this work, we propose to apply unsupervised clustering to mobile applications on their system call distributions. This is done by first adopting a static binary analysis that reverses engineering on executable of apps to find method call/sequence counts that are embedded in apps. Apps are then clustered iteratively based on this information to reveal implicit relationships among apps based on function call similarity. The GHSOM (Growing Hierarchical Self-Organizing Map), an unsupervised learning tool, is integrated to cluster apps based on the information resolved from their executable directly. We use types of methods and sequences as features. To run the clustering algorithm on apps, however, we immediately confront a problem that we have a large amount of attributes and data that leads to a long/infeasible analysis time with GHSOMs. The new iterative approach is proposed to conquer this problem along with dimension reduction with principle component analysis, cutting attributes with limited information loss. In the preliminary result on analyzing hundreds of apps that are directly downloaded from Apple app store, we can find that the proposed clustering works well and reveals some interesting information. Apps that are developed by the same company are clustered in the same group. Apps that have similar behaviors, e.g., having the same functions on games, painting, socializing, are clustered together.

WHO的一項全球性調查顯示，全世界真正健康的人佔5%，經診斷有疾病的人佔20%，處於亞健康的人佔75%，伴隨著慢性病所造成的死亡率也跟著增加，生活的品質已成為大眾們所關注的焦點，因此國人已有如何自主管理健康取代經常性的藥物治療，以增進健康的生活品質之意識。隨著國人對於自我健康管理的認知，如何滿足有效的管理健康需求的增加，已是大家探討的熱門議題。針對此一議題，本研究以健康促進處方為核心，整合個人化的生理健康數據如血壓、血糖、膽固醇、生活習慣、工作型態等進行健康風險評估，進而推薦適合的個人化健康促進處方，搭配運動器材、行動應用程式(Mobile Application)、雲端運算平台、體適能中心之運動生理監控，未來可適用於企業自行發展出的一套「雲端個人化健康管理平台」，有效支持個人運動安全並培養定期的運動習慣，達到健康促進的目標。

各式應用領域常會面臨許多撮合(Matching)問題，但當我們有需求時卻往往無法定出好的撮合策略，更遑論找到可實現此策略的電腦化解決方法。本研究希望針對穩定婚姻配對、大學聯考分發、論文審查分配、專題選修等等之類的撮合問題提供各種可行的通用撮合策略，可供使用者依其需求快速選用。而後續提供的支援系統則可據此產生一個以WEB為基礎的客製化專門應用領域撮合系統。 而什麼是撮合呢? 撮合是指有A、B兩群對象，在特定的規則與限制條件下，希望使每一A(B) 群對象可以連結至某些B(A)群對象，而使總體滿意度達到最大。以數學而言，一個A、B兩群間的撮合，就是一個滿足特定條件的A、B兩個集合間的二元關係。撮合類型可能是一對一、一對多、 多對多三種。一對一表示一個A群成員只能跟一個B群成員配對，一對多表示 一個A(B) 群成員能跟多個B(A) 群成員配對，多對多則指一個A群成員能跟多個B群成員配對且一個B群成員也能跟多個A群成員配對。 由於撮合型態與策略具有相當大的分歧性，以專用演算法實做並不實際，因此我們採用ASP(Answer set programming)實做撮合程式。ASP 是一種邏輯編程語言，具有宣告式程式特性，廣泛用於組合性問題的解決上，極適合應用在撮合策略的制定與實做。 在可真正執行撮合程式之前，必須預先建置A、B兩群對象的基本資料，因此我們的系統將允許開發者輸入A、B兩群對象的基本後設資料及撮合策略，而系統將據此建立對應Web介面與資料庫，允許使用者建立撮合對象的基本資料。一旦基本資料建立完成，系統即可依據系統設定的撮合策略以及以ASP實做的基本配對規則快速產生撮合結果，提供給使用者參考。 / There are a lot of application domains in which we may encounter the problem of finding a matching among two parties of entities. However, it is often the case that once a matching is needed, we cannot easily find a good matching strategy suitable for our purpose, not to mention one with a computerized implementation. This thesis aims to provide a web-based matching generation system allowing the quick generation of customized matching systems for users' need after their input of different demands of matching types and strategies. The supported types of matchings include most often used cases such as marriage/dating matching, paper review assignment, college admission dispatch and student-advisor selection etc. What is a matching? A (bipartite) matching problem contains two parties of entities, each member of which has a preference over members of the opposite party. A matching in a matching problem is a binary relation between both parties of entities. The goal of a matching problem is to find one or more optimal matching in which the total satisfaction of both party members is maximized. Matching problems can be classified according restrictions imposed on matchings. 1-1 matching requires each member of both parties to be matched to at most one opposite party member, 1-m matching allows only members of one party to be matched to more than one opposite party member, and m-m matching allows members of both parties to be matched to more than one opposite party member. Because there is a great variety of matching types and strategies, it is impractical to employ dedicated algorithm per case. It is thus eagerly expected to have a general framework in which different types of matching and strategies can be encoded. By applying Answer-set Programming (ASP) we provided one such framework in this thesis. ASP is logic programming language with declarative characteristics, widely applied in the solution of hard combinatorial problems, to be used in the encoding and solving of matching problems with different preference matching strategies. Theoretical discussion of matching algorithms always assumes that party members and their preferences are available in advance. However, to engineer a matching system, we still need to provide means to achieve it. Our system is thus also a matching support system, through the web interface of which developers and end-users can enter meta and individual information about all concerned properties and/or preferences of party members. After a possibly further processing of users' preference on the values of concerned properties of opposite party members for deriving every member's preference on the member of the opposite party, succeeding matching thus can obtain all needed data.

Kyoto University (京都大学) / 0048 / 新制・課程博士 / 博士(工学) / 甲第15332号 / 工博第3211号 / 新制||工||1484(附属図書館) / 27810 / 京都大学大学院工学研究科分子工学専攻 / (主査)教授 榊 茂好, 教授 田中 一義, 教授 北川 進 / 学位規則第4条第1項該当

科学研究費補助金 研究種目:総合研究(A) 課題番号:04302008 研究代表者:三井 斌友 研究期間:1992-1994年度

Despite the built-in garbage collector, Java programs can still suffer the memory leak problem resulted from the unhealthy programming style of retaining unwanted references. This paper presents an aspect-based tool for assisting programmers in locating such references to fix the problem. This tool, FindLeaks, utilizes an aspect to collect memory consumption statistics and object references created during a program’s execution and analyzes them for detecting memory leaks. The distinctive feature of FindLeaks is that it reports not only suspected classes of leaked objects but also where in the source the unwanted references were created. Besides, this paper also reports our experience with FindLeaks on three open source Java programs and how we enhanced it to lower its overhead.

在現在的社會裡，手機應用程式已經被人們接受與廣泛地利用，然而目前市面上的手機 App 推薦系統，多以使用者實際使用與回報作為參考，若有惡意行為軟體，在使用者介面後竊取使用者資料，這些推薦系統是難以查知其行為的，因此我們提出了 AppReco，一套可以系統化的推薦 iOS App 的推薦系統，而且不需要使用者去實際操作、執行 App。 整個分析流程包括三個步驟：(1) 透過無監督式學習法的隱含狄利克雷分布(Latent Dirichlet Allocation, LDA)做出主題模型，再使用增長層級式自我組織映射圖(Growing Hierarchical Self-Organizing Map, GHSOM)進行分群。(2)使用靜態分析程式碼，去找出其應用程式所執行的行為。(3)透過我們的評分公式對於這些 App，進行評分。 在分群 App 方面，AppReco 使用這些應用程式的官方敘述來進行分群，讓擁有類似屬性的手機應用程式群聚在一起；在檢視 App 方面，AppReco 透過靜態分析這些 App 的程式碼，來計算其使用行為的多寡；在推薦 App 方面，AppReco 分析類似屬性的 App 與其執行的行為，最後推薦使用者使用較少敏感行為(如使用廣告、使用個人資料、使用社群軟體開發包等)的 App。 而本研究使用在 Apple App Store 上面數千個在各個類別中的前兩百名 App 做為我們的實驗資料集來進行實驗。 / Mobile applications have been widely used in life and become dominant software applications nowadays. However there are lack of systematic recommendation systems that can be leveraged in advance without users’ evaluations. We present AppReco, a systematic recommendation system of iOS mobile applications that can evaluate mobile applications without executions. AppReco evaluates apps that have similar interests with static binary analysis, revealing their behaviors according to the embedded functions in the executable. The analysis consists of three stages: (1) unsupervised learning on app descriptions with Latent Dirichlet Allocation for topic discovery and Growing Hierarchical Self-organizing Maps for hierarchical clustering, (2) static binary analysis on executables to discover embedded system calls and (3) ranking common-topic applications from their matched behavior patterns. To find apps that have similar interests, AppReco discovers (unsupervised) topics in official descriptions and clusters apps that have common topics as similar-interest apps. To evaluate apps, AppReco adopts static binary analysis on their executables to count invoked system calls and reveal embedded functions. To recommend apps, AppReco analyzes similar-interest apps with their behaviors of executables, and recommend apps that have less sensitive behaviors such as commercial advertisements, privacy information access, and internet connections, to users. We report our analysis against thousands of iOS apps in the Apple app store including most of the listed top 200 applications in each category.

本文主要在考慮某類二階橢圓型偏微分方程式解的不存在問題，共分為三部分。 在第一部分中，首先利用均值函數之方法（此法曾見於〔Ｎ〕及〔Ｃ〕等多篇文獻 ）來研究如下之二階積分－微分方程式， ╭ (0.1) △u ＝ K(x)h(u)＋H(│x│)│ a(│y│)q(u(y))dy ╯R□ ，x R□,n ＞ 2 ，在此 ▔ (1) △表示ｎ維的拉氏(Laplace) 運算子。 (2) K(．),H(．) 及 a(．) 是局部赫德 (Holder) 連續的非負函數。 σ δ (3) h(．) 及 q(．) 滿足適當的條件，如 〞ｈ(ｕ) ＝ ｕ 及ｑ(ｕ)＝ｕ ” δｕ σｕ 或 ”ｈ(ｕ)＝ｅ 及 ｑ(ｕ)＝ｅ ” 。 當 H(．)≡0 時，鄭國順教授及林震燦教授〔Ｃ〕，已証明當γ足夠大時，若存在 某一正常數Ｃ，使得 __ Ｃ Ｋ(γ) ＞ ── ▔ γ□ __ （在此Ｋ表示函數Ｋ之均值函數），則方程式(0.1) 在Ｒ□中不存在任何正〔有界 〕的解。 令我們感興趣的是當 H(．)≡0 時，在那些條件下會有類似的結果發生。本文證明 ，當γ足夠大時，若存在某正常數Ｃ使得 __ ╭ ∞ n-1 Ｃ Ｋ(γ) ＋ Ｈ(γ) │ a(ρ)ρ dρ ＞ ─── ╯γ ▔ γ□' 則可經由詹森氏(Jensen's)不等式及赫德不等式，利用反證法去得到類似的結果。 在第二部份中，將研究下列之擬線性微分方程式解的不存在問題， ╭ (0.2) ．[g(│ u│) u]＝K(│x│)h(u)＋H(│x│)│ a(│y│) ╯R□ q(u(y))dy, x R□,n ＞ 2 ，在此 ▔ (1) u 表示ｕ的梯度。 __ (2) ｇ:Ｒ ──→Ｒ 屬於 Ｃ〔0,p□〕∩Ｃ□（0,p□）,p□ 為區間〔0,∞〕 □ □ 中之某一常數。 (3) (pg(p))'＞0 對所有的p (0,p□)。 (4) K(．),H(．) 及a(．) 均為局部赫德 (Holder)-連續的非負函數。 σ δ (5) h(．)及 g(．) 滿足適當的條件，如”h(u)＝u 及 q(u)＝u 〞 或〞h(u)＝ σu δu e 及 q(u)＝e ”。 首先定義函數ψ如下 ψ＝pg(│p│) p R. 若ψ的反函數存在，則可經由赫德不等式推導出一積分不等式，接著可利用此不等 式經由反證法得到下列的結果： (Ⅰ)在下列條件下 (a) 0 ＜ g(p) ＜ kp□ ，對任意非負常數ｍ及正常數ｋ以及所有p＞0均成立 。▔ ▔ (b) ｍ及ｎ滿足〞m＞0 且 n＞2〞 或〞m＝0 且 n＞ 3〞。 ▔ ▔ Ｃ (c) 當γ足夠大時，存在正常數Ｃ，使得 K(r) ＞ ──── 成立。 ▔ γm+2 ，當 H(．)≡0 時，方程式(0.2) 在 R□中不存在正〔有界〕的放射性解。 (Ⅱ)如果 g(．),K(．),H(．)及a(．) 滿足 (a) 對任意正常數ｋ及所有實數 p, 0 ＜ g(p) ＜ k ＜ ∞ 恆成立。 ▔ ▔ (b) 在γ足夠大時，存在正常數Ｃ使得 ╭ ∞ n-1 Ｃ K(γ)＋H(γ) │ a(ρ)ρ dρ ＞ ─── ╯γ ▔ γ□ ，則當 H(．)≡0 時，方程式(0.2) 在 R□中不存在正〔有界〕的放射性解。 在第三部份中，主要在處理如下之擬線性微分方程式的正放射解之不存在問題， ╭ │ ．[g(│ u│) u]＝f(│x│,u) x Ω, (0.3) ＜ u(x)＝0 x Ω, │ u(x) 0 x Ω, ╰ 在此Ω為 R□中之一球。 在[NT],[NM] 及 [NS] 中，作者利用波氏 (Pohozaev) 不等式去證明方程式(0.3) 在ｆ只含變數ｕ時，解的不存在結果。 在本文，將去探討當ｆ含變數ｕ及ｒ時，在何種條件下會使方程式(0.3) 不存在正 放射性解。首先，經由假設方程式(0.3) 存在正放射性解，吾人得到一個一般化的 波氏不等式，然後將其應用於部份擬線性橢圓型偏微分方程式上（如拉氏運算子， 平均曲率運算子及一般化平均曲率運算子），並去證明這些方程式在Ω上不存在任 何正放射性解。