近幾年來,全球資訊網在網際網路上蓬勃發展,因此而產生在網際網路上尋找資訊的問題。以目前的搜尋引擎與目錄服務的作法並無法找到真正有用的、潛在的資訊與知識。而有待運用資料發掘的技術於其上。但是,構成全球資訊網網頁之超文字遠比傳統的關連式資料庫非結構化。本研究先對資料發掘,全球資訊網資料搜尋等文獻與現況進行整理分析,並採用OCLC與NCSA所建議Dublin Core或Desai(1997)之Semantic Header為基本之超文字文件結構,再針對選定之應用領域(英語教育)建議彈性擴充之屬性。採用Han(1995)之概念樹及多層次資料庫的觀念,提出一可行之全球資訊網上資料發掘的整體架構;並發展一包含行業領域規則之資料發掘系統雛形,以驗証其可行性,並評估其日後實務操作之複雜性。 In recent years, a number of applications on World Wide Web (WWW) by Internet have grown rapidly. A unresolved problem of finding information on Internet has emerged. It is difficult for current search engines or directory services to identify valid, novel potentially useful patterns in data. Someone might hope to apply the techniques of data mining (or called knowledge discovery). However, the web pages written in HTML (Hyper Text Markup Language) are more unstructured than traditional relational database. This research began with literature review of data mining and WWW data searching and also evaluated the current business practice. The Dublin Core proposed by American OCLC and NCSA (1996), or semantic header proposed by Desai (1997) was modified as our HTML basic structure elements. This research also suggested some augmented attributes in a selected application domain (the English education). The concept hierarchy and multiple layered database suggested by Han (1995) will be adopted. A feasible integrated architecture of industry domain rule was then developed to test the feasibility and evaluate the complexion of this kind of system in the future real world.