基於PageRank與聲譽模型之社群書籤雲端平台系統

Abstract

近年來網際網路的蓬勃發展，人們學習新知識的方式不再侷限於書本上，隨著各大主流的搜尋引擎崛起，已成為現代人們分享、學習知識的一項重要管道。然而網路資訊日新月益，資料量愈趨龐大，但目前各大主流的搜尋引擎卻是單使用者的設計模式，導致無法提供人們進行協同合作過濾龐大的資訊。 本研究之主要目的是藉由基於雲端運算所開發的社會性書籤(Social Bookmark)網站─MyBook及IE Toolbar，協助人們進行協同合作與網站篩選，進而分享、學習知識。本研究利用使用者的書籤資訊，計算使用者彼此間的興趣相似程度，作為網站推薦機制及交友機制，並結合Google PageRank與聲譽模式(Reputation-based Model)，排序其網站與好友的推薦結果。因此本研究的主要貢獻為提供基於雲端運算所開發的協同合作平台，並結合現有的搜尋引擎，達到社群交友、協同合作及知識分享，除了能結識興趣相同的其他使用者之外，還能透過他人的經驗，在短時間內獲得所需之資訊。 本研究最後透過為期一星期的系統實測，蒐集52位使用者的行為資訊，並從中探討相似度、聲譽值、PageRank與引用次數彼此間之關係、使用者角色及社會性書籤系統設計等相關議題。

The development of Internet is thriving in recent years. People don't confined learn the new knowledge to books. After search engines appear, people can easily learn and search the new knowledge. Unfortunately amount of data is getting huge in the internet, many search engines do not explicitly support collaboration during search and this let people can not collaborate to filter data. This paper introduces two systems that social bookmarking website based on cloud computing and IE Toolbar. These systems support users by harnessing the bookmark's metadata、users' similarity calculation、PageRank and Reputation Model as the base for websites and users recommendation. Therefore, a key contribution of this paper is to develop a collaboration system and to design IE Toolbar to work with mainstream search engines. Let user can easily make friend with another user, collaborate to filter data and share their knowledge. Finally, the experiment invites 52 users and they use this system about one week. The research findings show that reputation, PageRank and citation times are positive relationship. Personality and interactivity can enrich the content and enlarge the user interesting in this system.