• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 36
  • 30
  • 22
  • 8
  • 6
  • 1
  • Tagged with
  • 73
  • 73
  • 14
  • 12
  • 12
  • 12
  • 12
  • 10
  • 9
  • 9
  • 9
  • 9
  • 8
  • 8
  • 8
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

中文新聞標題自動生成之研究 / A Study on the Automatic Generation for Headlines of Chinese News Articles

江珮翎, Chiang, Pei-ling Unknown Date (has links)
在網路資訊爆炸的年代,資料的分析整理日趨重要,本論文之研究目標正是針對資料做標題生成的處理,為資料自動生成標題,進而將資料加值化,轉化為資訊。研究者首先閱讀英文相關文獻,分析整理後,認為中文的處理方式與英文有所差異,因此,在本論文中,提出與英文不同之中文前置作業與自動標題生成之方法。 研究者針對標題的自動生成提出了幾種特徵值考量,包括候選詞權重值,訓練標題-文本詞彙,標題長度的關係及詞組間距。本論文之研究分為兩階段,第一階段為訓練階段,將文件做前置處理與斷詞,接著訓練標題-文本詞彙與統計文件標題長度的機率。第二階段為執行階段,分析新文件之候選詞權重值,並參照訓練階段之標題-文本詞彙與標題長度之機率值參考表,考量詞組間距後自動為文件產生標題。本論文所採用的訓練文件集來源為1998年至1999年五種報紙,涵蓋不同主題,共84,211篇文件,而測試文件的實驗分為Outside Test與Inside Test兩部分。 研究者為實驗結果進行兩種評估,一為電腦評估,將自動生成之標題與記者所擬訂的標題比對後,計算出求準率、求全率與F1。Outside Test求準率為14.21%、求全率為11.43%、F1為12.67%。Inside Test求準率為15.84%、求全率為12.94%、F1為14.21%。實驗結果顯示,正確率方面與其他文獻之英文文件標題的生成結果(F1=3.2%~24%)相近,但與實際標題仍有差距,因此,在未來工作上,仍有很大的發展空間。二為人為評估,讓使用者在閱讀自動生成之標題後,加以評分。自動生成之標題的流暢度還算不錯。然總結來說,本論文之研究尚屬初始階段,盼未來能更加成熟,並可有更進一步的創新與改進。 / As the number of digital documents on internet is growing up, analysis and organization of documents become quite important. In this thesis, we propose an approach for headline generation of documents. We can try our best to transfer the document data into information in some sense using the proposed approach. We review literature about the related topics, and present a different approach to deal with Chinese documents rather than English documents. We propose some approach to Chinese documents headline generation. The thesis is separate two steps, one is training step, and the other is execution step. On the first step, the documents were preprocessed. Secondly, we trained the probability of headline-text words, and headline’s length. And on the execution step, we analyzed scores of headline candidates and gap, then referred to the probability of headline-text words, and headline’s length, finally we automatically generate headline for documents. The training documents are selected from a test collection for information retrieval, CIRB. Totally 84,211 Chinese news articles published between 1998 and 1999 are selected. Testing documents has two parts, one is for outside test, and the other is for inside test. We conducted two evaluations, one is the automatic evaluation using metrics of presicion, recall and F1; the other is the human assessment. The precision of outside test is 14.21%、recall is 11.43%、F1 is 12.67%. And the precision of inside test is 15.84%、recall is 12.94%、F1 is 14.21%。The automatic evaluation result shows the accruacy is still not good enough, and the human assessment evaluation shows our approach can produce human-readable headlines.
2

熱帶幾何之圓錐與凸集的生成元素探討 / The Generators Of Cone And Convex Set In Tropical Geometry

詹佑民 Unknown Date (has links)
此篇論文我們主要是探討熱帶幾何下,凸集(convex set)以及圓錐(cone)的生成元素(generator)個數。在第二章中我們對一些基本環境及運算工具做介紹,例如:熱帶半環(tropical semiring)為度量空間(metric space)、極限值的運算性質等等,在第三章中我們探討回收錐(recession cone)、凸集及圓錐的性質,其中包含三者之間的關係,而在第四章中我們探討閉圓錐(closed cone)、緊緻凸集(compact convex set)、閉凸集(closed convex set)三者的生成元素個數,並以實例說明此性質,最後我們將推論出一個方法來找出在二維的熱帶空間底下的有限生成圓錐之生成元素。 / In this thesis, I will discuss the generators of cone and convex set in tropical geometry. In Chapter 2, basic environment in tropical geometry and arithmetic tools are introduced here, such as how to find the limit value in tropical geometry or deciding if tropical semiring is a metric space, etc. In Chapter 3, I explore the properties of cone, convex and recession cone, inclusive of the relations of one another. In Chapter 4, the generators of a closed cone, a compact convex set, a closed convex set are provided with illustrations to present the properties. It will finally lead to a method to find the generators of the finitely generated cone in two dimesion space.
3

有關k元數列的探討 / A Study about k-Sequences

江玲慧 Unknown Date (has links)
本篇論文主要探討長度為n的k元數列,其中若有i個偶數,j個奇數, 個不限制其奇偶,符合條件的數列個數。第一章 先預備後面計算所需要的基本知識,第二章 由生成函數開始推導公式,第三章 再討論 時的特殊情況,並利用組合方法來加以證明。第四章 針對生成函數推導出的公式再深入探討。第五章 檢討與展望。
4

相對財富與貨幣內生成長:最適化的不完全競爭總體模型 / Relative Wealth And Monetary Endogenous Growth:The Optimizing Model With Imperfect Competiton

蕭淑云, Hsiao,Shu-Yun Unknown Date (has links)
本文建構一個商品市場為不完全競爭以及經濟體系中存在相對財富效果的貨幣內生成長模型,並採用消費受到現金付現的方式將貨幣引入模型中,而商品市場為不完全競爭的特質乃是根據Benhabib and Farmer (1994)的設定,並討論在這個架構之下,經濟體系的相關動態特質。 由於體系中存在著相對財富效果,因此得到貨幣成長率增加會提振經濟成長率的結果,此乃是不同於既存文獻之處,所以可清楚的了解到相對財富效果所扮演的重要角色。接著分析預料到恆久性貨幣成長率與預料到恆久性資本所得稅稅率對體系中的動態路徑所造成的影響。 再者,因為體系中存在相對財富效果及商品市場為不完全競爭這兩種特質,使得分權經濟的競爭均衡並未達到社會的最適,有鑑於此,本文最後一章乃採用Greiner and Hanusch (1998)和Shieh, Lai and Chang (2002)極大化社會福利的處理方式,找出體系中最適的貨幣政策與最適的財政政策,且不同於既存文獻之處在於,本文會藉由所求出的最適政策來矯正體系中所存在的扭曲,以使得體系回至社會的最適境界。 最後,本文得出經濟體系中存在相對財富效果以及商品市場為不完全競爭這種特質時,則央行與政府均可以找出最適的貨幣政策與最適的財政政策,分別來矯正體系中的扭曲現象。
5

Catalan數的對射證明 / A Bijective Proof of Catalan Number

李英杰, Lee, Ing-Jye Unknown Date (has links)
本文的主旨是利用對射函數的方法,證明圓周上2n個點成功配對問題的解是Catalan數.所以必須找一個也是Catalan數的事物來和本問題對應,這裡找的是n個節點的二元數.我們先造一個由成功配對應射到二元數的函數,再證明此函數是一對一且映成,既為對射函數,則我們就可以知道成功配對的解是Catalan數.然後再將問題推廣到3n個點,甚至到kn個點的情形,以得到一般的問題解.
6

在點對點網路上針對串流資料傳播的品質保證 / Quality assurance of streaming data dissemination over p2p network

邱威中, Chiu, Wei Chung Unknown Date (has links)
網路技術發展的日新月異帶領了眾多新網路服務的崛起,例如即時影音串流這類的多媒體服務。但即時影音串流服務所產生的龐大資料流和傳輸延遲時間的嚴格限制也隨之而來的為網路環境帶來許多挑戰,在這些條件下,傳統Server-client拓樸架構將client要求的影音資料以單一鏈結傳輸時,常會因為頻寬不足而面臨嚴重的封包遺失,或是資料流擁擠造成的額外傳輸延遲使得封包無法達到即時性的需求。P2P網路擁有server-client架構所難以達到的規模伸縮性,且對於節點、鏈結失效所引起的傳輸錯誤也較能容忍,更重要的是,它有效的分散了原本負載在少數link上的龐大資料流。因此P2P架構近年來風行於即時影音串流服務。 目前P2P網路的拓樸多是隨意形成,當網路成員規模龐大時,由傳送端出發到遠方的接收端,途中可能經過無數的鏈結,每一個鏈結都會由於頻寬的不足使得資料流遭受某種程度的品質損害,另一方面,對即時影音服務而言,若資料流的累積延遲時間超出可容忍範圍時,無法為使用者接受。 本研究嘗試找出一個較好的拓樸用以傳輸多媒體資料流,使得位於最遠端節點的累積延遲亦能為使用者接受,且資料品質的損害程度最小。我們將之建置成一NP-Complete複雜度的問題模型,名為MLDST。而解法則是修改Dijkstra single-source shortest-path演算法,並加上每個節點承擔下游節點數量及延遲時間限制而來。我們以PlanetLab環境在實際的網路上進行實驗,證實我們的演算法比傳統的Minimum-Spanning Tree及shortest path spanning tree有更好的影像品質。 / Numerous new network services arise with the advanced development of network technologies, such as real-time multimedia streaming services. But challenges to network environment come along with the enormous traffic of data flows and rigorous restriction to transmission delay of real-time multimedia streaming services. Under this circumstance, conventional server-client topology suffers from serious packet loss and packet delay due to the overload of servers and their accessing links. Also, extra transmission delay may make packets fail to meet the requirement of real-timed services. Peer-to-peer network is more scalable than server-client model, and is much more tolerable to the transmission errors caused by node or link failures. More importantly, it effectively distributes load from the server to peers. As a consequence, peer-to-peer service architecture becomes very popular for real-time multimedia streaming services recently. Peer-to-peer networks are mostly formed in random fashion. As the size of network grows, packets may have to travel through numerous links to reach far-end receivers. The quality of data may be damaged by insufficient bandwidth of links. For real-time multimedia services, it is not acceptable to users if the cumulated packet delay exceeds a tolerable limit. Our research is trying to find a better topology to transmit multimedia data flows which makes the cumulated delay of the most-far-end user be tolerable and the damage of data quality is minimized. The problem is modeled as a MLDST problem, which is a NP-Complete problem. To solve the problem, we modified Dijkstra’s single-source shortest-path algorithm by bounding the node degree and adding delay constraint. The experiments were carried out on real network environment through PlanetLab. Experiments show that our algorithm outperforms traditional MST and shortest path spanning tree.
7

基於L-system之動態模擬東方風格立體雲紋生成技術 / L-system base Dynamic Simulation for 3D Oriental Cloud Pattern

郭明諺, Ming-Yen Kuo Unknown Date (has links)
本篇論文對於傳統的東方風格雲紋進行分析,歸納其中的生成規則與排列特性,並將雲紋圖樣作簡單的分類,發現雲紋圖樣具有自我相似與遞迴組成的特性,此點特性非常符合原本用於模擬植物生長的L-system,所以我們就針對所蒐集的眾多雲紋圖樣參考資料所分析出的資訊定義了一個適用於立體東方風格雲紋的生成生長規則,以L-system之技術為基礎進行撰寫規則語法,利用所定義結構化之規則循序地在三維空間中動態模擬雲紋的生長與變化。更進一步地,加入具時間序列的參數調整,可以使得雲紋隨著時間軸的變化產生東方風格立體動態雲紋。 / Cloud Pattern is an important and common element in oriental decoration art, it is a challenge to extend the 2D abstract line into 3D objects. In this thesis, we try to induce the pattern generation rule and the feature of oriental cloud by analyzing the ancient oriental cloud pattern. We make classification from the oriental 2D cloud pattern, and discover the pattern is combined with symmetric, self-similar and repetitive features. These characteristic are the core of L-System, which is a suitable framework for plant growth and simulation, so we generate a new rule with specific parameter from the feature of 2D cloud pattern to generate oriental cloud pattern in 3-dimention space, simulating the growth of cloud and variety in structure. Further, we add the time temporal parameter to control the 3D oriental cloud pattern generation process dynamically.
8

プログラム生成系GeneSysにおける等式仕様への否定の導入

NISHIDA, Naoki, KUSAKARI, Keiichirou, SAKABE, Toshiki, SAKAI, Masahiko, KONDO, Satoru, 西田, 直樹, 草刈, 圭一郎, 坂部, 俊樹, 酒井, 正彦, 近藤, 悟 12 1900 (has links)
No description available.
9

条件付き等式の変換に基づくプログラム生成

KUSAKARI Keiichirou, NISHIDA, Naoki, SAKABE, Toshiki, SAKAI, Masahiro, NAGASHIMA, Masanori, 草刈, 圭一朗, 西田, 直樹, 坂部, 俊樹, 酒井, 正彦, 長島, 正憲 12 1900 (has links)
No description available.
10

二酸化チタンナノチューブ光触媒を用いた水素生成・分離一体型メンブレンの開発

服部, 真史 26 March 2012 (has links)
Kyoto University (京都大学) / 0048 / 新制・課程博士 / 博士(工学) / 甲第16803号 / 工博第3524号 / 新制||工||1533(附属図書館) / 29478 / 京都大学大学院工学研究科電子工学専攻 / (主査)教授 松重 和美, 教授 藤田 静雄, 准教授 後藤 康仁 / 学位規則第4条第1項該当

Page generated in 0.0338 seconds