2015-02-012024-05-17https://scholars.lib.ntu.edu.tw/handle/123456789/678339摘要：營建產業具有高風險的特性，使得營建工程安全的確保及工程意外災害的防範成為產業內最迫切需要解決的問題。然而，營建工程專案與製造業不同，每個專案皆有其獨特性，使得工安意外的發生情境皆不盡相同，也提升了制訂標準防災流程的難度，諸如潛在危險因子難以辨識、災害防範流程難以符合現況而形同虛設…等等。在營建產業當中，最易取得的營建工程安全知識素材仍是以文本資料（諸如營建工安規範、重大傷亡報告等等）為主。雖然分析純文字資料的技術亦相當的多樣與成熟，諸如資訊檢索（Information Retrieval）及文本分類（Text Classification）等等，但是純文字資料雖然適合人閱讀與理解，其對於語意或知識內涵的表達仍然有其極限，並且由於其未經結構化的特性會導致電腦判讀上的誤解，也因此在知識擷取與整理的自動化上存在較多的障礙。 本研究擬以文本探勘（Text Mining）技術為基礎，提出一將未結構化的文字資料逐步整理為結構化的資料庫的流程，使其成為電腦可解讀（Machine Readable）的資料，並繼而導入更多的資料分析技術（如資料探勘，Data Mining）從中整理出營建工程安全領域的潛在知識，俾以從事更多後續之可能應用及資訊系統之開發。 <br> Abstract: Construction industry is surrounded by high potential occupational hazards. It makes construction safety management an important and urgent issue. However, due to the uniqueness of the construction projects, the experiences of violation preventions cannot be duplicated without any change. It increases the difficulty on developing the safety regulations and standard operation procedures for the construction safety management. Within the construction industry, the most available resources are text units such as safety standards and fatality reports. Information Retrieval and Text Classification are matured technologies that can manage and access the text units. However, the text units are unstructured or semi-structured that are not machine readable. It increase the difficulties for the automation of knowledge discover. Based on Text Mining techniques, this research project proposes a procedure that transforms the text units into data base format, and therefore make the text resources structured and machine readable. After that, Data Mining analysis is also applied to discover the implicit knowledge within the data base. The outputs of the knowledge discovering process will enable more possible applications for the construction safety managements.文本探勘資料探勘知識發掘營建工程安全Text MiningData MiningKnowledge DiscoveryConstruction Safety