在一般內容管理系統中,使用者常常無法找到想要或相關的數位內容資源, 主要原因是在比對或是推理的過程中,無法獲得足夠描述這些數位內容資源的語 意資訊,作為運算的對象。 本論文提出內容模型(content model)來描述數位內容資源,並透過自動化的語 意描述機制來建立數位內容的內容模型檔案,以節省人力與時間的消耗,透過以 本體論文基礎的自動分類機制,提昇資訊檢索準確度,並達到個人化知識的再利 用與分享。為了提升資訊檢索準確度的效率,我們也整合LSI (Latent Semantic Indexing)技術與個人註記工具,以提昇描述數位內容資源所需的語意資訊,這些 語意資訊會存到內容模型資料庫。 透過ICU 內容管理系統(Content Management System)實作環境以及後續的資 訊檢索實驗,我們驗證了內容模型所提供的語意描述能力,能夠確實提昇檢索的 性能。實驗結果顯示,我們提出的在自動化擷取內容模型的機制,能夠截取出大 部分的語意資訊,因而可以有效的節省建立語意資訊的人力與成本。在內容管理 系統中,個人化註記知識能夠有效的與被註記的數位內容資源結合,達到知識的 分享與再利用,並能增進資訊檢索的性能。 Users usually may not find desired content via existing Content Management Systems. The main reason caused this problem is the lacking of enough semantic description for digital content during the analyzing, reasoning, or computing process. In this thesis, we proposed the Content Model to provide the semantic description of digital content. A computer-aided description mechanism is developed to assist user establishing content model to reduce the needed cost and laboring. To improved the performance of information retrieval, we proposed ontology-based classification mechanism combined with Latent Semantic Index method to enhance the semantic relation of Content Model, which represented digital content. To demonstrate the proposed semantic description, we implement the content model via ICU Content Management System. The experiments results indicate the proposed Content Model can improve the retrieval performance. The computer-aid description mechanism can extract great part of meaningful information and does reduce the required cost of Content Model establishments. Besides, in Content Management System, the personalized annotation can be combined with described digital content to achieve knowledge reuse and share.