以網站日誌探勘建立網站架構

Using Web Log Mining to Create a Website Structure

洪范文(Hung Fan Wen)

指導教授 ： 謝建成

國立臺灣師範大學/教育學院/圖書資訊學研究所/碩士(2010年)

摘要

在各種資訊被大量數位化的網路環境中，網站已變成各機構所必須建立的資訊據點。使用者是否能快速並且直覺地找到需要的資訊，是網站提供給使用者搜尋所需資訊的能力，亦即為一網站內部的尋獲度。
由於網站結構的日趨複雜，使用者往往需花費較多的時間才能找到所需的資訊，如能根據使用者的瀏覽行為修正網站的結構，可使網站具有適應性，進而提昇使用者的瀏覽效率。但目前有關網站架構的研究、或是局部調整網頁結構等，大多未能從整體的觀點最佳化網站的結構。因此本研究提出以目標網頁的因素加入傳統的方法中，根據網頁間的瀏覽次數，以權重的大小決定某一網頁的父網頁進而重整網站的結構，以提高網站內部尋獲度。
本研究提出一演算法計算瀏覽序列之時間門檻與觀察使用者瀏覽行為找出目標網頁，以決定該網頁的所有父候選網頁之權重，藉此重建網站架構，並以任務導向之尋獲度測試成效。測試結果顯示目標網頁次數比重為100%的架構，其架構之尋獲度成效較高且優於現行架構，亦優於傳統只考慮網頁瀏覽次數之架構，進而建構符合使用者行為之網站架構。且目標網頁次數比重的不同所組織的架構也明顯不同，在組織結構上，本研究所建構以目標網頁次數比重為100%的架構是屬於廣而淺的形式，在尋獲度成效上也確實能符合前人研究的結果。相較於研究個案中的現行架構，其廣度稍微減少，更可減少使用者在搜尋項目時的迷失感，較符合一個良好資訊架構的設計方式。該架構雖然在部分任務上表現不盡理想，但整體上有著比原先架構與傳統方法之架構有著較佳的表現。

關鍵字

尋獲度 ； 網站日誌探勘 ； 資訊架構

並列摘要

In the internet environment where all types of information are digitized, a website has become the information site which all institutes and organizations must establish. Whether users can rapidly and intuitively find out information they need is the ability that a website provide for them to do it; i.e. it is the internal findability of a website.
Because website structures become increasingly complicated, users often need more time to find out needed information; if website structures can be amended according to users’ browsing behaviors, then they can be adaptive and it can improve users’ browsing efficiency. However, most of the current studies related to website structure or partly-adjusted website structure cannot optimize website structure from a holistic viewpoint. Therefore, this study proposed adding the factor of target page into the traditional method, determining the parent page of a certain page by weighing the browsing times of these pages and further reorganized website structure based on it to improve the internal findability.
This study proposed an algorithm to calculate the time threshold of user session, observe users’ browsing behaviors and find out the target page to determine the weight of all parent page candidates which was the basis to rebuild the website structure, and tested the outcome by the task-oriented findability. The results showed that the findability of the structure whose target page frequency weight was 100% was higher than the current one and than the traditional structure which only considered page browsing times, and this study further constructed a website structure conforming to user behavior. Moreover, the organization structure was significantly different as the frequency weight of target page was different; as for the organization structure, the structure that the frequency weight of target page was 100% constructed in this study was a broad and shallow form, and the result of findability did answer to the results of former studies. Compared with the current structure in the cases, its breadth was lightly lowered and this could reduce users’ lost feeling during information seeking and was a better design of information architecture. Although this structure did not perform well in some parts of task, it did have a better performance than the former and traditional ones as a whole.

並列關鍵字

Findability ； Web log mining ； Information architecture

參考文獻

張碧顯（民97）。從網站日誌中發現網站結構及網頁關係。未出版之碩士論文，國立暨南國際大學，南投縣。

蘇俊斌（民94）。應用網站探勘技術於網友瀏灠行為分析-以內容服務網站為例。未出版之碩士論文，國立台灣大學，台北市。

丁依玲（民96）。以卡片分類法分析大學圖書館網站架構－以國立臺灣師範大學圖書館網站為例。未出版之碩士論文，國立台灣師範大學，台北市。

邱涵芬（民95）。應用資料探勘與文件分類建構個人化知識查詢網站─以全民健康保險研究資料庫與醫藥新聞為例。未出版之碩士倫文，私立大同大學，台北市。

賴玲玲、鍾閔衛（民96）。建構資訊服務Ⅱ：以網頁設計人員為例。教育資料與圖書館學，45(1)，33-53。