Intelligent e-Learning Recommendation and Navigation

碩士論文[[abstract]]網際網路的便利與時效性，改變了現代人的生活習慣。資訊的快速傳遞與更新，讓人們不再滿足於傳統的教學方式。儘管網路教學的研究越來越多，但能根據個人的喜好去推薦適合自己課程之相關研究，以及將資料探勘技術運用在網路教學之研究中，甚少人著墨。 有鑑於此，本研究除了利用關鍵字與同義字的概念，同時依據領域知識以一般化與特殊化之原理來導航適合學習者難易程度的課程相關內容，並利用精確的個人化推薦讓學習者能學習的更有興趣及更有效益。本研究之主要方法，是利用關聯式法則之資料探勘技術與階層分析法結合，發展出策略式權重之資料探勘技術，亦即根據學習者的喜好程度，利用階層分析法給予附加權重的配置。該附加權重主要目的是避免傳統資料探勘技術在支持度設定太低時挖出太多的規則，設定太高時挖出太少的規則，可能無法挖出新的喜好規則而喪失了較好及適當的推薦機會。所以本研究用階層分析法求得學習者有興趣的課程類別之附加權重後，再探勘出學習者的實際學習行為。此方法結合學習者主觀的喜好以及客觀的行為概念來將學習者聚類，進而作為課程推薦的依據。如此，除了能達到課程導航之目的，更能精確且快速地推薦符合學習者興趣之相關課程。 最後將本研究方法與傳統的資料探勘技術作一比較，並以本校大學部的學生作為系統驗證的調查對象，藉以證明本研究所提出的策略式資料探勘可以更有效率且精確地推薦學習者想要學習的課程內容，進而達到智慧型網路教學課程推薦與導航之目的。[[abstract]]Convenient and speedy Internet changes our lives and habits. As fast-updated information, people are not satisfied with the traditional teaching method any more. Few researches according to personal preferences recommend courses and few researches use Data Mining in e-Learning .Thus, this article uses the concept of the synonyms and the key words to find courses. The navigation is suitable for courses according to generalization and specialization of ontology, and let learners have interests and effects by personal recommendation precisely. The main method is Strategic Data Mining that integrates Association Rule and Analytical Hierarchy Process. According to learners’ preferences with Analytical Hierarchy Process, we add weight to avoid traditional Data Mining rules that can’t mine new preferences and loss opportunities for better and suitable recommendation. Therefore, we use Analytical Hierarchy Process to acquire weight that learners’ interesting courses and then mine learners’ practical behavior. Integrate their subjective preferences and objective behavior, and then turn into the recommendation of base courses. In those ways, we can reach the navigational purpose and recommend precisely the courses those learners’ interests.In the end, we compare Strategic Data Mining with traditional Data Mining. Students of our university are the research objects. We will prove Strategic Data Mining that can recommend courses precisely and effectively. In those ways, we can reach the navigation and recommendation purposes