臺灣博碩士論文加值系統

隨著資訊科技的發展與網際網路的盛行，網路上所提供的豐富資料，已可部份滿足使用者對於各類知識與服務的需求。然而資訊檢索技術的進步，雖可協助資訊需求者便利地尋找資料，但卻可能造成「資訊超載」與「不適當資訊充斥」的問題有鑑於此，本論文即嘗試提出一種結合現有「搜尋引擎檢索功能」的智慧型代理器系統架構。首先，本研究將網路上網頁，依照其功能區分為文章摘要、研討會資訊、學校課程資訊、書籍介紹、常問問題集、新聞、個人或團體網頁、公司宣傳與產品廣告等八種網頁資訊。透過隸屬度函數的設定、模糊法則的制訂及推論過程，以完成網頁的辨識工作。經由網路上隨機選用的811張網頁分類的實驗結果顯示，利用所建置的資訊過濾智慧代理器，可在兼顧效率和滿足個人特殊需求下，大幅簡化人類篩選資料的程序，故可視為一種值得開發、並加以商業化的網路工具軟體。

Due to the breakthrough of computer technology and of the developmentof Internet, it is possible to find out lots of information availablefor network users. However, the abuse of information retrieval toolshas the side effects of information overload and inappropriateness.To solve those problems, a novel intelligent agent is proposed byincorporating with the current search engines of Internet.First of all, web-page information is functionally classified intointo eight categories: article summaries, seminar minutes, courseschedules, book instructions, FAQ, news, individual or group homepages, and company promotions and product advertisements.Based on the formulation and inference of Fuzzy rules, the proposedsystem can successfully recognize the categories of tested webpages.Experiments of classifying 811 web-pages show that the system canboth improve the efficiency of information searching and match thespecial demands of users. Therefore, the system is worth to befurther developed and commercialized.