涉毒刑事案件实体关系语料库构建OA

Publicly available structured textual data on drug-related criminal cases remain scarce,hindering data-driven research in this domain.To bridge this gap,this study collects drug-related criminal judgments from China Judgments Online and constructs a structured corpus with fine-grained entity and relation annotations.The annotation schema covers entities such as drug types,involved individuals,and behavioral patterns,as well as semantic relations including drug trafficking,possession,and illegal harboring.We establish a detailed annotation protocol and ensure data quality through an iterative multi-round labeling process with cross-validation.Statistical analysis reveals class imbalance in the annotated corpus:entity types such as personal names occur frequently,whereas certain relations(e.g.,possession)are relatively sparse.The resulting corpus provides high-quality foundational data for downstream natural language processing tasks in drug crime analysis,including named entity recognition,relation extraction,and intelligent early-warning systems.