This study proposes strategies to automate public record reclassification tasks and
enhance reliability using generative AI and large language models (LLMs) in the digital
transformation era of public records management. Reclassification, a critical
a...
This study proposes strategies to automate public record reclassification tasks and
enhance reliability using generative AI and large language models (LLMs) in the digital
transformation era of public records management. Reclassification, a critical
administrative process re-evaluating previously non-public records for disclosure
potential, currently depends heavily on manual efforts, resulting in inefficiencies and
error risks.
To address these challenges, the research designs a service model that automates
sensitive information detection, document classification, and de-identification by
integrating natural language processing (NLP), generative AI, and Human-in-the-loop
methodologies, while reflecting legal-institutional environments and security frameworks.
Domestic and international case analyses alongside pilot project outcomes confirm AI's
superiority over manual methods in processing speed, sensitive information detection
accuracy, and decision consistency.
Results demonstrate that generative AI, when integrated with records management
systems, substantially alleviates burdens in screening, analysis, and judgment phases for
mass electronic records, while Human-in-the-loop structures preserve legal- 65
accountability and quality control. User interfaces and explainability further elevate user
trust and acceptance. Ultimately, this study offers a conceptual architecture and
implementation procedures for a generative AI-based reclassification service, providing
policy and technical directions for digital innovation in public records management and
strengthened accountability and transparency in information disclosure; future research
should pursue institutional pilots, quantitative evaluations, and domain-specific model
development.