- 學位論文
基於Meta搜尋和資料探勘技術的自動化網路資源推薦系統
Automatic Web Resource Recommendation System Based on Meta Search and Data Mining
摘要
由於資訊科技的發達，網際網路的使用率越來越高，因此，網路已然成為了快速且及時獲得知識的平台。然而，網路上充斥著各式各樣的資訊，造成了多數使用者無法快速的找到自己需要的資訊，即使透過搜尋引擎找尋資訊，其龐大的資訊卻讓人無所適從，使得整理資訊的過程消耗過多的人力。在本論文中所研究的「教育」領域，由於其科目、主題及課程架構與概念太過於複雜，老師對於不同的課程，很難有效率的找到適合用於教學的網路資源。因此，本論文提出以Meta Search及Web Data Mining技術，提供教材推薦服務，讓老師在授課時，能夠更快速的取得用於教學的數位教材。首先，藉由領域專家 (domain expert) 提供與領域相關的領域關鍵字 (domain keywords) 讓系統從網路上收集與領域相關的網路資源，接著從這些網路資源中擷取metadata並累積到資料庫中，再透過片語擷取 (phrase extraction) 與關鍵字權重分析 (keyword weighting) 技術將這些網路資源分別整合並推薦到不同的課程單元中。最後我們將以這個系統提供服務平台，以群眾外包 (crowdsourcing) 概念不斷改進系統推薦網路資源的正確性。
並列摘要
With the development of information technology, the Internet becomes most important information platform on the world. However, too much information make people cannot find the useful information efficiently. Even if we can get the information from search engine service, the search engine can still not provide specific domain content exactly. It costs too much manual effort for searching domain-related web resources. In the case of education domain, the architecture and concept of the subjects, topics and lessons are too complex. Teacher cannot find the useful material for each lesson. In this thesis, we apply Meta search and web data mining technique to automatically recommend digital material. Let teachers get the digital material which can be appropriate on teaching. In the first phase, the system would collect the domain-related web resources from the Internet by the domain keywords supported by domain expert. By extracting the metadata in these web resources, the system accumulates information semi-automatically. Then, the web resources would be integrated and recommended to different lessons by phrase extraction and keyword weighting technique. We would support a service platform and improve the precision of the recommended web resources based on concept of crowdsourcing.