隨著數據時代的迅速發展，網路以及文字資訊的大數據爆量增加。世界各國政府機構、民營企業以及各研究團隊相當重視大數據分析，並希望能夠從這些巨量資訊中，擷取最相關的資訊，以做為決策參考。行政院飛航安全委員會自1998年成立後，以累積相當之飛機失事調查報告並公布於網站上。本研究擬以文字探勘方式，對於92件飛航失事調查報告，擇定內容之四個分析結果分別為可能肇因、與風險有關、其他調查發現，與提供飛安改善建議，透過文章斷詞與文字雲視覺化呈現，並區分為動詞與名詞，以呈現飛安事故發生之安全趨勢。研究發現，名詞出現頻率最多者為駕駛員、組員及跑道，動詞部分以進場、落地及操作為最多，顯示飛機失事之安全趨勢，以進場及落地階段，因飛行組員之相關操作有關。

With the fast development of the internet era, the explosion of big data on the Internet and text mining is increasing accordingly. Government agencies, domestic or international cooperations, and research institutes worldwide have emphasized the importance of data mining analysis, and hope to extract the most relevant information from these huge amounts of information for decision-making reference.

Since the establishment of the Aviation Safety Committee (ASC) of the Executive Yuan in 1998, it has accumulated a considerable amount of investigation reports on aircraft crashes and published on the website. Current study analyzes four analytic results for 92 flight accident investigation reports, including possible causes, risk-related, other investigation findings, and safety recommendations, and intends to use text mining technology- PowerBI- to invest the trend of aircraft accidental contributing factors. This paper intend to present the contributing factors, catagorizing into verbs and nouns, visually through word segmentation and word cloud, and analyzes the safety trend of aircraft accidental reports. Conclusions of current study indicates the safety trend of aircraft crashes are contributing to the inappropreiately operations of the flight crew in approach and landing stages since the most frequent nouns are pilot, crewmembers and runway, and the verbs are approach, landing and operation.

KEYWORDS： Big Data Analysis; Text Mining; Accident Investigation; Power BI