Research on Dynamic Judicial Hallucination Risk Identification Mechanism for Legal Domain Large Language Models
This study constructs a dynamic risk identification mechanism based on verifiable legal corpora collected from public statutes, judicial interpretations, judicial documents, and typical cases that transforms legal hallucination from an opaque generation error into a computable, explainable, and reviewable judicial risk...