Geographical network analysis of drug trafficking in China (2012–2024): A method based on large language models

This article analyzes the geographical spatial model of drug trafficking crimes (domestic transportation in China and cross-border transportation involving China) from 2012 to 2024. Based on 8,357 cases of drug trafficking crimes, key information on drug transportation (transportation routes, transportation methods, etc.) was extracted using large language models (LLMs). For the two main drugs, methamphetamine and heroin, transportation routes at three levels (national level, provincial level, and city level) were constructed. Through social network analysis, the patterns of drug trafficking crimes in China at different scales were revealed: (1) The macro-pattern of these two drugs mainly shows transportation from the two southern provinces of Guangdong and Yunnan to northern provinces. (2) the southwestern region of China, including Yunnan province, Sichuan province, and the major municipalities of Chongqing are high-frequency areas for drug trafficking crimes, and the route ”Myanmar-Yunnan-Chengdu/Chongqing-other provinces” is an important drug transportation path. (3) Drug trafficking crimes in China are concentrated in economically developed regional central cities. Based on these findings, targeted control suggestions for different types of drug trafficking crimes were proposed. The method of combining LLMs with social network analysis proposed in this paper can be applied to the analysis of other crimes such as smuggling.