在本論文中,我們設計一個搜尋相關網頁代理人系統,幫助使用者在網路英語學習的環境中,快速找到可供學習閱讀的網路文章,提高學習的效率。 在網頁蒐集部份,不同於一般搜尋引擎蒐集方式,我們只蒐集學習者曾經瀏覽過網頁與其延伸的相關網頁,節省了網路資源及儲存空間的花費。在搜尋相關網頁部份,當學習者想要找出目前閱讀文章的相關網頁時,為了避免在對語言還不熟悉的情況下,輸入不適當的關鍵字,浪費許多找尋相關網頁的時間,所以我們設計了不需輸入任何關鍵字,就能自動找出這篇文章相關網頁的方法,讓使用者很快的找到更多相似文章來閱讀。 在實作中,我們利用代理人系統個人化的特性,來蒐集使用者曾經瀏覽過的網頁主題,並根據相關網頁會有相互連結的特性,將同一主題的網頁抓取回來,這部份利用到Google搜尋引擎工具,幫助我們取得網頁的父連結。在搜尋相關網頁方面,我們採用連結為基礎的搜尋方式,再結合內容為基礎的概念,提高相關網頁內容的權重,讓找到的相關網頁更加符合學習者閱讀的需求。 In this thesis, we have designed an agent that aids English learners in finding related web pages that would help them in learning and thus increase their learning proficiency. The fact that many users not fluent in English may provide inappropriate keywords and end up with unrelated search results is acknowledged. Unlike common search engines, we have designed an agent in which web pages related to an essay can be found easily and quickly for the learner, without ever having to provide a keyword. Our system can be personalized according to each user. It first recognizes the browsed web page’s theme, finds related links, and finally obtains other web pages with the same theme. To search for related web pages, we first used links as a primary search method, and then combined this search with each webpage’s content. This would help in finding more relevant web pages, as the importance of the content is acknowledged.