本研究的主要目的在描繪相較於印刷報紙的情形下，網路BBS使用者現階段對電子報的認知圖像，並且探討這些認知圖像與網路BBS使用者的閱報行為及網路使用行為是否具有關聯性？若有，則關聯性為何？ 本研究以交大資科BBS站的註冊使用者為對象，以等距抽樣的方式從中抽出500位樣本，利用電子郵寄問卷的方式進行調查。研究問卷中主要列出電子報與印刷報可諮比較的題項，以試圖從受訪者的回答中，整理歸納出他們在相較於印刷報的情形之下，對電子報的認知及看法為何。此次調查約有效回收問卷總計有276份。 研究結果發現，在受訪者的認知中，電子報和印刷報最大的不同在於「互動性」、「易取得／易瀏覽」及「發行對象」等特質因素上。受訪者認為，在媒體與證者的互動功能上，電子報明顯優於印刷報；不過，印刷報紙則相當容易取得，無論何時何地都可閱讀，而不像閱讀電子報具有地域（一定要有電腦的地方才行）上的限制。 此外，受訪者也認為，印刷報的發行對象是沒有限制的，只要讀者其有基本的識字能力，即可毫無困難的閱讀；相反的，閱讀電子報除了識字能力外，還須具有基本的電腦能力才行。 至於在受訪者的閱報行為及網路使用行為與電子報認知之關聯性分析中，本研究初步發現，受訪者閱讀電子報的頻率和其對電子報「新聞可信度」的認知有顯著正相關(r=.22)；但與其對「電子報內容」的認知則呈負相關(r=-.22)。另一方面，受訪者的網路使用行為與其對電子報的認知則無太大關聯性。雖然在關聯性分析上，不顯著的情形居多，但值得注意的是，呈顯著相關的面向都表現在「電子報的內容」此一因素上，而且是負相關。這意味著，目前接觸電子報或網路愈頻繁的閱聽人，對電子報內容的評價卻愈低，因此，如何改進電子報的內容以吸引更多的讀者，實在值得研發者深思。 本研究還發現，受訪者中非電子報讀者對電子報的認知評價高於電子報讀者，高評價或許正因為他們對電子報持有較高的期望，而如何使電子報在網際網路上呈現的型態能符合閱聽人的期望，也是值得電子報業者注意的。 / The main purpose of this research is to portray the internet (BBS) users' cognitive intage on electronic newspaper at this stage, compared with that on printed newspaper. Besides, this study will try to discover if there is a relation between the BBS users' reading behavior on newspaper and utilizing behavior on internet. If there is one, we would like to know what it is. This research takes as subjectes the registered users of CIS BBS at NCTU , selecting 500 samples in the way of interval sampling. Listed in the research questionnaire are the comparable items available between printed and electronic newspapers. We try to find out, in the answers of the subjects, their cognition and impression on electronic newspaper when compared with printed newspaper. The outcome of the research tells us that in the cognition of the subjects, the major differences between printed and electronic newspaper are factors such as interactivity, availability/browsability, and objects to publish for. The subjects recognize that electronic newspaper is obviously superior to its printed counterpart in terms of interactivity between media and the readers. The printed newspaper, however, is better in terms of availability, ready to be read at anytime and anywhere; while reading electronic newspaper has to be confined within the locations where there are computers. Furthermore, the subjects also think that the printed newspaper is published for everyone--as long as they are literate, they can read without difficulty; on the other hand, to read electronic newspaper the readers must have basic command in operating computer in addition to being literate. In the correlation analysis of the subjects' reading behavior on newspaper and utilizing behavior on internet with their cognition on electronic newspaper, we find initiatively in the research that the subjects' reading frequency on electronic newspaper and the cognition of 'reliability of news' on electronic newspaper have a saliently positive correlation (r=.22), but the frequency and the cognition of 'content of electronic newspaper' have a negative correlation (r=-.22). On the other hand, the correlation between the subjects' utilizing behavior on internet and their cognition on electronic newspaper is very low. Although in the correlation analysis non-salient cases prevails, but it is worth noticing that all the salient correlation aspects appear on the 'content of electronic newspaper'--and the correlation are all negative. This means that the more frequently the readers access electronic newspaper or network, the lower their evaluation toward the content of electronic newspaper. Therefore, it is a serious problem for the developers to ponder about how to promote the quality of the content of electronic newspaper in order to attract more readers. We also find that non-electronic newspaper readers' evaluation on electronic newspaper is higher than that of the electronic newspaper readers. Higher evaluation perhaps reveals their higher expectation to electronic newspaper. How to make the way electronic newspaper presented on Internet meets the expectation of the readers is also a problem for the publishers fo think about.

由於現在網路發展的非常快速，資料的產生速度以及使用方式已經超過人們的分析能力與解讀能力，因此近來大資料計算是一個很重要的研究課題，許多科學家與研究單位都積極地投入大量的研究資源，目的就是要研究這些龐大的資料要如何去分析，或是解讀。特別是大型網站上的資料，使用者的數量和可點選的項目是隨著時間不斷增加，如何分析這類的資料是一個重要的課題。 我們將介紹如何利用Python程式的特性對大型網站進行使用者偏好分析，透過亂數投影和分解-合成多元尺度法的合作，做到使用者偏好網路的建制，協助大型網站對使用者進行即時性的閱讀項目推薦。我們提出Data is cache for dimension reduction 的概念，說明大資料計算必須配合資料庫才能達到真正的快速計算結果。