線上百科全書的計量比較研究-以Wikipedia與Knol的高品質文章為例

Abstract

本研究企圖以量化方式比較兩種不同類型的線上百科全書，共擬訂了三個重要的研究步驟，包含選定研究對象、建立比較指標以及建構抽樣原則。首先必須選擇研究對象，即Wikipedia與Knol，且分析彼此的特色與差異；第二是建立比較指標，根據系統資訊與文獻回顧發展六大指標，為每年平均瀏覽量、正文字數、網頁可讀性、每年平均被引數、引文數量以及引文類型，從中比較議題熱門程度、內容多寡、適合閱讀的年齡層、網路世界的可見度、引用多少參考資源建構文章內容以及引用資源之廣度（比例）與深度（評比）等面向；其三是建構兩大抽樣原則，為抽樣高品質以及探討相同議題的文章，據此可得20個議題。由於抽樣結果是屬於成對小樣本，較適合採用無母數統計的魏克森符號等級，並根據六大指標比較兩線上百科全書在相似文章之異同，作為改善系統與建置文章的依據。 研究結果顯示Wikipedia與Knol在每年平均瀏覽量、每年平均被引數、引文數量與期刊影響係數有顯著差異；正文字數、網頁可讀性、期刊百分比、書百分比、網路資源百分比、書評與網頁排序無顯著差異。 本研究歸納以下幾點結論。Wikipedia在每年平均瀏覽量、每年平均被引數、引文數量上的數據較高，但任何人都可編輯文章內容，使用者必須小心使用，培養自身辯證資訊真實性的能力，避免吸收錯誤資訊。Knol雖已訂定首選文章之評選指標，但著重在瀏覽量、閱評以及評論，導致8篇文章沒有引文，有待系統發展相關管理規範，敦促文章補齊參考文獻資訊。兩線上百科全書的網頁可讀性都是12，相較於目前美國成年人的閱讀能力，還是稍嫌過高；兩者亦尚未設定服務的目標讀者，可斟酌使用對象調整文章的難易程度。

This study attempts to compare online encyclopedias in a quantitative approach, and conducts three important research designs. One is choosing research targets, that is, Wikipedia and Knol. After deciding the encyclopedis, researcher analyzes and presents their differences. The second step is developing 6 indices, including page views per year, text words, readability, cited numbers per year, citation numbers, and citation types. These indices represent the popularity of the topics, content richness, readability levels, use how much sources to write articles, and width (percentage) and depth (assess) of the citations, respectively. The third step is setting up two main sampling rules, which sample the same topics and choose the high quality articles. According to the sampling rules, totally 20 topics are selected. Finally, the researcher compares the differences using the Wilcoxon signed-rank test, suggests the ways to improve systems and construct articles. The research results show that page views per year, cited numbers per year, citation numbers, and impact factor are statistically significant based on confidence level at 0.95, but the remaining indices are not. The major findings are shown as follows. When it comes to page views per year and cited numbers per year, values of Wikipedia are higher. However, Wikipedia can be written by everyone, users must use it cautiously and need to cultivate the ability of identifying the authenticity of information to avoid absorbing wrong messages. Although Knol has already set up an assessment criterion, which emphasizes pageviews, reviews and comments, for selecting the best articles, and still has 8 articles without citations. Knol should take it as a serious problem and force to set up related standards. Besides, the readability of Wikipedia and Knol is 12. Due to American adult literacy survey, the readability of these topics is too high to read. Two encyclopedias have not yet set service objects which can depend on to adjust the degree of readability levels.