Research on Crime Investigation Verification Methods through Natural Language Processing-Based Analysis of Court Judgments

The recent amendments to the Korean Criminal Procedure Act have brought significant changes to the investigative environment of the police. By granting the authority to conclude investigations to the police, the process of reviewing cases by police investigators has become unprecedentedly crucial. Furthermore, the growing trend of direct investigation of evidence in courts demands logically complete arguments in the courtroom. Consequently, comprehensive and well-structured reasoning for both investigation and verification of cases is anticipated to become a vital competency for the police. While efforts are being made to ensure objectivity in investigative outcomes, the constraints of the investigative environment make it difficult for investigators to allocate sufficient time and resources to analytical investigations. However, despite these limitations, it is imperative to establish a process of verifying investigative results based on evidentiary analysis and logical reasoning, considering the constraints faced by law enforcement agencies. In light of this perspective, this study aims to propose a systematic approach to investigative verification by utilizing primary trial judgments, which share a similar structure with criminal investigative reports. By applying artificial intelligence-based natural language processing techniques, this research postulates that even with limited personnel resources, prompt and efficient analysis of cases can be achieved, leading to improved completeness of investigations. To explore this, the study conducted experiments on automatically extracting argument structures from trial judgments used as data. Based on the results of these experiments, the research endeavors to design a scientific argument structure for crime investigation verification, ensuring the legitimacy and objectivity of investigations through analyzing legal reasoning structure present in trial judgments. The findings of this study are expected to provide a foundation for advancing research in the field of natural language processing in the domain of crime investigations.