運用文本探勘建置九二一地震數位檔案知識庫

Using Text Mining to Build Digital Archives Knowledge Base of Chi-Chi Earthquake

蔣以仁(Chiang, I-Jen)

知識庫 ； 文本探勘 ； 智慧資產 ； 圖形辨識 ； 特徵萃取 ； knowledge base ； text mining ； intellectual property ； pattern recognition ； feature extraction

檔案季刊

8卷3期（2009 / 09 / 01）

44 - 66

繁體中文

經驗累積了知識，創造了價值；政府的施政作為與歷程，皆記載在於公文檔案中，而這些寶貴的資料，如何開啟，成為決策施政者的參考，遂為當前檔案管理進到檔案知識管理上重要課題。檔案管理局於民國93年對九二一地震政府文書檔案資料庫進行知識探索作業，利用文本探勘機制，以九二一地震政府公文為標的，從關鍵詞彙萃取、自動化的分類分群，用以發掘出對災難的處置，行政作為的經驗等之內隱知識，建構知識庫，以達到「提供國家發展見證，創造國家智慧資產」。

Experiences create valuable knowledge. The experiences of the public affairs are recorded in official records that formally reflect the activities of a government. It is an important topic of National Archives Administration to discover the hidden knowledge in records for governors' references, while they consider making a decision on related public affairs. In 2004, National Archives Administration started to use text-mining techniques in order to discover the knowledge behind the whole collection of Chi-Chi earthquake official records that is able to provide enough information for emergency rescue and to properly handle the aftermath of the accident. This project went on through key phrases extractions, automatic text classifications/clustering, etc., to create the ontology of disasters control, especially such as earthquakes. The aim of building knowledge by using text mining is as the slogan - ＂To provide evidence of progress and to create the intellectual property of our country＂ at National Archives Administration.