OCR與機器學習在檔案內容辨識之初探

The Initial Exploration of OCR and Machine Learning in Archival Content Identification

國家發展委員會檔案管理局自成立迄今，積極徵集國家檔案，辦理檔案數位化以永續典藏珍貴檔案，並期將相關成果運用於遠距服務及研究，發揮各種加值價值。目前使用者可透過國家檔案資訊系統查詢目錄資訊，使用全文影像，但仍希冀能以資訊技術提供更多檔案內容服務，爰本研究蒐整光學辨識技術應用，挖掘深化國家檔案內容之可能性。

National Archives Administration, National Development Council (hereinafter referred as NAA) has been acquiring and digitizing national archives since its establishment. And NAA is looking forward to make use of the above results for remote services and researches to utilize various added values. Currently, users are able to search national archives catalog via Archive Access service and use archival images. Still, NAA would like to provide more content services by trying out new information technologies. Thus, this research combed the development and implements of current optical character recognition (OCR) technologies to explore possibilities to uncover more archival content.