- 學位論文
以語意網為基礎的企業資訊入口網站資源整合與個人化
Semantic Web-based Information Process and Personalization Applying to Enterprise Information Portal Resources Integration
摘要
企業資訊入口網站(Enterprise Information Portal, EIP)主要是運用Web 的技術，將不同來源的資訊整合在單一的介面，讓企業內外不同使用者，皆能使用此單一入口進行資訊管理。然而EIP將不同資訊系統功能集中在一起使用，提供使用者單一Web使用介面，即使功能已經集中起來，但在使用上還是要一個一個個別地使用，對於進一步內容上的自動整合卻沒有學者提出相應對策。自全球資訊網之父Tim Berners-Lee於西元2001年提出語意網(Semantic Web)概念後，Web內容自動整合的研究成為一股潮流，也為EIP各獨立功能內容資源整合調配自動化的研究注入新希望。 現階段語意網的研究皆集中在知識塑模與網路資源的內部註記上，也就是說是採用由下而上(Bottom up)地建構語意網知識庫，以提供自動化的資源整合機制，EIP內容整合的研究也多半集中在「原生」地產生知識本體描述，所謂的「原生」即為其本來就提供知識本體予語意網的代理程式(Agent)進行分析，如果EIP沒辦法提供其自我資源描述的知識本體，語意網的代理程式即無從進行分析與內容的理解。 因此，本研究提出一以外部註記為基礎的EIP資源整合與個人化框架，應用結構驅動的網頁資源擷取器自動從EIP各應用功能模組中提取資料，再利用欄位樹外部註記方式將資料轉換成具語意描述的知識實體，匯入語意網知識庫與個人知識本體進行整合，進而達到自動從現有基礎資源提取出語意網知識庫的目的，之後便可依據特定需求對知識庫進行推理查詢，找出間接或是需要推導的隱性整合資訊，即達到自動化整理並整合特定資料的目的。另外，經整合後的個人化資訊，其資料原本是存在於EIP各功能區塊中，一但從原應用中抽出(經過知識庫的查詢)，轉化為整合資訊，很可能會讓抽出的資訊與原應用失去連結，使得我們無法迅速得知某部分整合資訊的來源，本研究利用資訊視覺化技術Cluster Map來輔助呈現整合資訊，其可以畫出整合資訊與原應用功能資源交集的部分，如此即可達到連結原功能資源與整合內容資源的目的。 我們通過案例模擬(Simulation)測試，證實本研究之成果有助於企業工作者提升其工作效率。提供web資源的自動化處理、內容整合與本體重用是目前語意網科技標準欲達到的願景，本研究基於以上理論與技術，提出了可供參考的解決方案。
並列摘要
Enterprise Information Portal (EIP) uses web technology to integrate different sources of information systems into a single user interface and makes enterprise employees and clients perform information management more easily. Although EIP puts diverse systems into homogeneous EIP modules, the contents of the modules are still isolated. For the integration of the contents and mining, exploring the contents of EIP, a further research is still needed. Since the Pioneer of World Wide Web, Tim Berners-Lee, introduced the concept of Semantic Web, the research of web resources automation processing and integration has been the trend. Additionally, it brings new hopes to the research of integration of EIP modules. At the present stage, the research of semantic web mostly focuses on knowledge modeling and internal annotation of the web resources. It utilizes Bottom up method to build a semantic web knowledge base in order to provide an automation mechanism of web resource integration. The research of EIP contents integration primarily focuses on generating ontology natively. In other words, it provides ontology for the semantic web agents originally. If there is no ontology schema of the EIP, the semantic web agents could not exploit the EIP contents. Therefore, we present an external annotation-based EIP resource integration and personalization framework to help the EIP, which lacks a self-description schema, benefit from semantic web technology. The framework uses structure-driven web crawler to retrieve the contents for each EIP module. Subsequently, it applies slot-tree filling algorithm to annotate the documents in a way to make them meaningful. At the same time, it transforms the meaningful documents into RDF instances and combines them with specific personal ontology. Consequently, this framework builds a semantic web knowledge base of the EIP in which it allows users to query and obtain integration information. Nonetheless, the integration information inevitably loses its context, and it leads users not to be able to find the original information comes from quickly. We utilize Cluster Map to display the integration information and the original context in an attempt to solve this problem. Finally, through a case study and simulation, the framework of this research successfully facilitates enterprise employees to work more efficiently. To provide the automation processing of web resources, web contents integration and reuse of ontology is the main tasks and challenges of semantic web technology. This research provides a suggestion solution with the theory and practice of the semantic web technology.