法律資料分析與文字探勘：跨境毒品流動要素與結構研究

台灣約有六成毒品來自境外，因此「跨境」毒品流動為重要研究主題，亦關乎台灣在國際政治上之非傳統安全領域，但至今少有公開資訊。本研究使用法律資料分析法，探索七萬餘篇法院判決書作為文本資料，描繪「毒品案件要素與結構」。研究發現：（1）近年約有三分之二件毒品案件是累犯，所有案件中之四分之三可易科罰金，兩者佐證現實上監獄負擔過重問題。（2）近年關於毒品之7種犯罪方式之數量增長程度有顯著差異存在，犯案數量由大至小為「施用、持有、製造、販賣、轉讓、運輸、栽種」。（3）就近年案件而言，三級毒品愷他命（Ketamine）和四級毒品含其原料跨境走私情形最為嚴重；不過在「施用、持有」上卻是二級毒品最多。（4）本文使用結構方程模型發現關鍵字中，「以物販運」對「毒品種類」呈現正向路徑效果，可以佐證跨境毒品多以進口物品為由夾帶回台。本文發現台灣跨境毒品流動主要結構為：「由犯罪團隊以貨物從大陸走私或夾藏方式運輸四級毒品入境，將其作為先驅原料，再製造出二級毒品轉而輸出或在國內使用」。由於此一結構具有「輸入／加工／輸出」的特性，本文指出台灣很可能已成為跨境毒品鏈之節點。本研究主要貢獻在於處理巨量文字資料的實做與法律資料分析法，可有效提取法院判決中可茲使用之資訊情報，可茲作為厚資料實踐之可能範例。 Illegal drug is a serious global problem today. It is necessary to understand its distribution and trafficking routes in order to tackle this problem. Moreover, it is estimated that approximately 60 percent of the illegal drugs in Taiwan came from overseas. Hence, the flaw of cross-border drug control is also an important issue in international affairs. This article uses legal analytics to study 71,629 judgements text involving drug-related crimes to grasp the picture of case factor and structure by text mining technology. It is found that two-thirds of the above-mentioned cases were recidivists, and three-fourths of the total cases were subject to a fine at first. Secondly, the type of drug-related crimes keeps changing during different time periods. The types of drug-related crimes from most common to least common are as follows: use, possession, production, sale, transfer, transportation, and cultivation. Thirdly, the third-level drug, Ketamine and the fourth-level drug including raw materials all came from overseas, but most of the drugs being used and possessed are the second-level drugs. Fourthly, by SEM（Structural Equation Modelling）, this article points out that the most frequently seen characteristics of cross-border drug flaw is that the sale of goods is positively correlated with the type of the drugs, which illustrates that the main drug transportation route is from cargo by air or sea or smuggling. The research also discovered the main structure of its flaw is that the raw materials, often categorized as the fourth-level drug, are smuggled from mainland China, and then manufactured locally to be second-level drugs for export to Japan or other developed countries. In other words, Taiwan has possibly become a node of drug production in the perspective of the global drug trade chain. Lastly, this research has successfully applied text-mining to the large amount of legal texts, which is undoubtedly Think Data, to retrieve useful information and shows that legal analytics is possible and practical nowadays.