透過您的圖書館登入
IP:18.224.214.215
  • 會議論文
  • OpenAccess

資料採礦應用於偵測漁船走私之研究

A Study on the Taiwan's Fishing Vessels Smuggling Detecting-Applied Data-Mining Technique

摘要


在國家安全與反恐議題的要項中,走私防範是一項重要的工作,走私行為不但可能將各項直接影響國家安全的人、物密秘運入國內,或私運出境,其所造成國家經濟上、防疫工作、生態保護上均會造成重大的影響,因此,走私防範已是國際及國內反恐、安全及保障各項經濟秩序與國民健康上,最受重視的工作。我國海岸巡防範圍總長達1,820公里,每日於台灣沿近海之漁船筏逾14萬艘次,因此漁港安全檢查任務在走私防範的工作扮演重要的角色。海岸巡防機關依據「海岸巡防法」,執行漁船筏(民)進出港安全檢查,並建立「安檢資訊系統」資料庫。本研究以「安檢資訊系統」資料庫內資料為研究對象,藉資料採礦技術建立漁船(民)走私預測模型,成功發現北部地區、中部地區、南部地區及東部地區,各具不同的走私規則,透過決策樹與倒傳遞類神經網路之混合模型得知,即可準確掌握96%左右的走私訊息。本研究建立之漁船(民)走私預測模型,可提供海巡機關勤務分派參考,大幅降低漁港安檢投入的各項成本,而達到事半功倍之效,對於漁港緝私工作而言將能提供重要的決策支援效果。

並列摘要


In the study fields of national security and anti-terrorist, the prevention of fishing vessels smuggling is one of the most important issues. The works of national economics, diseases management and ecological protection will be impacted by vessels smuggling behaviors. Therefore, prevention smuggling is the most important work in international. For implementing the fishing vessels security check, according to Coast Guard Act, Coast Guard Administration has set up ”security clearance information system” to manage the fishing vessels data. The database is including all basic information, daily record of port traffic, and past smuggling records for each fishing vessel. In this study, we adopt the data mining techniques to analysis the data of ”security clearance information system” from 2008 to 2009, a total of 3583570 records of fishing vessels operation data. Through the decision tree, artificial neural network and both combined hybrid data mining skills, there are 96% smuggling information could be detected. It is an important research results for the prevention smuggling missions.

延伸閱讀