「科技訊息分群圖譜」導入政府決策支援系統之應用

掌握全球各國科技發展狀態是國家科技政策制定的重要工作，其成功關鍵就在於如何取得、呈現、處理相關資訊以提供決策者使用。然而目前國內外皆缺乏這類政府實務研究。因此，本研究提出政府科技政策的資訊蒐集、呈現、分析的重要性，並引入「國際科技政策觀測系統」，它是典型的政府科技決策支援體系，能有效蒐集資訊、減少決策環境的不確定性，這種環境偵測系統可作為環境預測與前瞻的基礎，對政策的學習與推動、科技的競爭與發展有正面助益。其次，我們透過文字探勘技術以及「增長階層式自組織映射圖」（growing hierarchical self-organizing map, GHSOM）群聚分析方法，將「國際科技政策觀測系統」中以文字型式呈現之文件資料，轉換為以圖像型式呈現的「科技訊息分群圖譜」（science & technology message grouping map, STMGM），建立「政策觀測圖像化模式」（policy patterns observed image, PPOI）應用，研究文字探勘結合圖譜之應用與潛在價值，以提升科技決策支援工作。本研究發現「科技訊息分群圖譜」、「政策觀測圖像化模式」在政策觀測工作上之應用，有助於提升決策者清楚、簡略、迅速了解國際科技政策發展概況的需求，完成強化系統的環境監視功能。這種將人類智能與資訊科技組合的應用，在科技政策領域將開啟新的決策支援模式，並啟發未來其他領域的政府決策支援工作思維。

關鍵字

文字探勘；資訊視覺化（圖像化、可視化）；環境偵測（環境掃描、環境監視）；決策支援系統；科技訊息分群圖譜

並列摘要

Getting hold of the development information in other nations is an important task in the process of national science and technology policy making. The key of success resides in how to obtain, process and present related information for decision makers. However, this kind of practical governmental work is hardly found in both local and foreign researches. This research emphasizes the importance of the collection, presentation, and analyses of related information for Science and Technology policy makers. It also introduces the International Science and Technology Policy Observing System (ISTPOS), a typical government science & technology decision support system, that aims at collecting information efficiently, whilst reducing the uncertainty in decision-making environment. As an environmental detection system, it acts as the basis of environmental prediction and science and technology foresight, greatly helping learning / promoting of policy and competition / development of such. Furthermore, we employ the cluster analysis of the Growing Hierarchical Self-Organizing Map (GHSOM) to illustrate the applicability of text mining and mapping on science & technology decision support task. This transforms document wording into graphic presentation of the Science & Technology Message Grouping Map (STMGM), to set up the Policy Patterns Observed Image (PPOI) applications. It in turn works on the application and potentiality of combining text mining and mapping to promote the decision support system. This work finds that by applying the STMGM and PPOI to policy observations, decision-makers can better understand the development/trends in international community, to strengthen the development of systematic environmental scanning. It is also encouraging to see the support approach, combing human intelligence and information of science and technology to adopt the PPOI in other fields of government decision support system in other fields in the near future.