As the time limits for opening archives become shorter, archive types become more complex and the number of archives continues to grow, it is necessary to explore automated methods for the open appraisal of archives. This improves the efficiency and professionalism of open appraisal, addresses the heavy workload and the large shortage of staff with the relevant professional skills, and better ensures that open appraisal work proceeds smoothly. Through research on the technical logic and application methods of AI-assisted open appraisal of archives, this project built an AI model for the open review of archives. Based on semantic learning technology, the system performs a comprehensive analysis of the full text of archives, combines it with external information such as the archive title and author, and processes sensitive information in the archives. It also combines traditional tools for restricting access to archives, such as keywords, to improve the accuracy of intelligent open review of archives.