Decoding the complexity of homemade explosives: Forensic insights from association rule mining on 344 Chinese Case...

Homemade explosives (HMEs) present significant challenges to forensic investigations due to their diverse chemical compositions and varying construction methods. Identifying the origin of these explosives is crucial for linking evidence across crime scenes. To address this challenge, this study employs an advanced data mining technique to enhance the forensic analysis of a unique dataset consisting of 344 HME samples collected from 129 real cases in China over an eight-year period (2015-2022). The samples were initially categorized based on their functional roles - main charge, initiator, or precursor/other - and analyzed using standardized forensic methods. Given the complexity of certain mixtures and the limitations of expert-based interpretation, Association Rule Mining (ARM) was subsequently employed to uncover hidden patterns in the chemical profiles. By interpreting the ARM findings, the study revealed valuable insights into the common "recipes" used by bomb-makers in China, identified frequently co-occurring components, and provided new forensic signatures that could potentially link different cases. Furthermore, the ARM-generated rules suggested likely HME ingredients and construction methods, even when only partial samples were recovered. To our knowledge, this study represents the most comprehensive forensic analysis of HMEs in China to date. It has great potential to contribute to international efforts in combating terrorism and other crimes involving the use of HMEs.