以語意網技術整合網站異質資料供鏈結資料發布

Using Semantic Web Technology to Integrate Heterogeneous Data on Websites for Publication of Linked Data

全球網從文件網提升至資料網的過程背後，語意網標準扮演主要的概念架構角色。語意網在現有異質資料層上，以對映、轉換、包裝等程序，透過共通的詞彙和資料模型建設語意資料層，在上層提供高階服務。本文採用過去成功的方法步驟，以國家發展委員會檔案管理局網站異質資料為研究目標，將上述概念具體實現以建立系統架構。網站資料分成結構性和文字資料，依循資料的準備、儲存及發布等步驟，按部就班在各階段探討如何引進標準技術及利用開放源工具，開發系統元件。對於結構性資料，研究如何在不影響現有資料庫系統運作的情況下，即時擷取關聯資料庫資料轉成RDF（Resource Description Framework）格式，提供鏈結資料發布之用。非結構性資料有賴於自然語言處理中的資訊擷取應用，將非結構性資料轉成鏈結資料。知識圖譜源自語意網／鏈結資料技術，提供鏈結資料發布以外的應用，本文也探討結合知識工程方法，設計相關知識系統，開發智慧型服務。

Semantic Web standards and technologies play the essential role of constructing conceptual architecture when enhancing the Web from document-based link to web of data. On the existing heterogeneous layer, we use procedures such as mapping, converting, and wrapping to build up the middle semantic data layer with common vocabulary and data models as the basis to provide higher-level services on the top layer. In this paper, we employ the successful method and steps, take the National Archives Administration website as the research goal and implement the above concepts to establish a system architecture. The resources on the web site are divided into structured data and textual data. Following the steps of data preparation, storing, and publishing, we investigate how to bring the standard technologies and leverage open-source tools to develop system components. For the part of structured data, we explore how to generate RDF data from relational databases for the publication of Linked Data without affecting the operation of the existing database system. For the part of textual data, we rely on the application of information extraction in natural language processing to convert textual data to Linked Data. Knowledge Graph originated from Semantic Web/Linked Data technologies provides applications other than Linked Data publication. In this paper, we also discuss the use of knowledge engineering methodology to design knowledge systems and develop intelligent services.