因應檔案法的相關規定，我國各政府機關陸續開始整理其檔 案，建立電子目錄，其中經鑑定具永久保存價值的檔案，移轉至檔案 中央主管機關，進行檔案全文影像掃瞄工作。此等數量龐大的政府檔 案電子目錄及部分全文數位化資料中，蘊藏著許多政府施政的智慧， 然而若任由此知識寶庫束之高閣，或僅提供一般性全文檢索查詢，至 為可惜。是以期能建構一套政府知識管理策略，運用電腦整理、分類、 研析功能，自政府檔案資訊系統裡，尋找知識，以創造檔案加值運用 價值，提供決策支援參考，強化行政工作效能。 由於美國檔案中央主管機關於文件管理自動化及自文件萃取知 識已有深入研究，居世界領先地位，本研究先先就其相關作為，進行 探討。此一研究發現，美國檔案中央主管機關發展知識管理係在電子 文件典藏策略計畫下，將檔案擷取知識的功能，列為必須涵括的研究 項目。其重點工作首先以建構知識為基礎的永久保存檔案系統架構為 首要核心項目。其次透過良好的系統架構，發展知識的探索與應用， 以因應日後可正確、有效地找出永久保存的電子檔案。規劃之主要精 神是將電子檔案資料與瞬息萬變的資訊科技軟硬體區隔開來，並自收 集資料時，便將擷取及運用檔案知識所需要的資訊涵括在內。故在儲 存電子檔案資料時，同時紀錄資料的產生、轉置、儲存等各階段之背 景資訊，以利未來應用時，得以依據該背景資訊存取所需檔案資料。 其系統架構分收集、儲存管理、使用三部分，每一部分在資料、資訊、 知識三層皆有其對應作為。整體架構採網格(GRID)技術，各類訊息溝 通以標準標示語言(tagging language)處理，透過反覆地收集、使用 程序，結合轉置作業，使資料得以繼續運用。經由循環作業，將可因 應科技變化及資料長久保存原則，惟應掌握標準標示語言及mark-up language 的變化。 由於我國檔案中央主管機關成立較晚，相關人力、經費、制度、 經驗雖不及美國，但因無負擔，可跳躍逕行規劃各項電子化作業方 式。自政府檔案資訊系統裡尋找知識的相關做法，可分長期及短期， 長期而言，以建構知識為基礎的永久保存檔案系統架構為必需規劃建 置之要項。短期而言，規劃建構政府知識管理作法，概分知識收集、 知識分類、儲存管理、知識傳播與分享、回饋方式五大部分，並分別 提出建議作法，以供實務界參考運用。 / Responding to the new movements of the Archives Act and related regulations, government agencies of ROC is beginning to archive their documents as well as records, and to setup corresponding electronic catalogs. Among others, the Archives with permanent preservation value are transferred to the National Archives Administration, and they are electronically stored. Thousands of government documents and records are cataloging and some of them are digitalized and stored into computer databases. Since a lot of wisdoms of national politics are inherited in the archives, it were a great pity to let this knowledge are stored merely in the archive chambers or only to be used for the purpose of general search for full-texts. It is very important to set a strategy to manage the knowledge gained from government’s documents. Moreover, it is important to use computers for checking, classifying, cataloging knowledge retrieval from archives management information system. These value-added archives are provided to decision makers for reference to increase administration efficiency. We choose the U.S. National Archives & Records Administration (NARA) as the survey target for its outstanding performance in electronic records management and knowledge retrieval. The most important result of this survey is that we have fund out that the NARA regards the knowledge retrieval from archives as an essential research topic in the knowledge management under the guidance of the plan of “Electronic Records Archives”. The primary job is to construct the infrastructure for the permanent storage of the knowledge-based records, so that this knowledge can be retrieved in the future correctly and effectively in means of this well constructed system, when the functions of exploration and the fields of usage of knowledge are further developed. The main idea is to separate the electronic records from the changing information software and hardware. Moreover, information about how to retrieve and use the records are analyzed and preserved with the archives. That is, detail background information for record generation, transformation, storage stages are kept for future use. The system architecture is consisted of three parts: ingest, storage management and access. Each part is mapping to data, information, and knowledge. The system uses grid technology, and all communication messages are coded in standard tagging languages. Through iterative ingesting, accessing and transforming, data could be used continuously. Iterative process copes with changing technology and abides by the principle of long term record-storage. The National Archives Administration is newly setup for ROC government. Compared to NARA, we haven’t reached the same level in workforces, financial supports, regulations, and experience. However, we have chances to work out electronic operations to utilize the quantum leap without much historical burdens. This paper carries out the long-term and short-term plans for archives management. In the long run, the priority goes to infrastructure for knowledge-based permanent archives storage. In the short term, this paper outlines the guidelines for government knowledge management, including knowledge collection, knowledge classification, storage management, knowledge sharing, and feedback to which may be useful to all stakeholders.