Scholarlink

In order to promote the intelligent protection of world cultural heritage, this paper focuses on the construction of local databases in its vertical AI knowledge infrastructure. Professional dispersion characteristics of cultural heritage data method and general adaptation shortage problem, put forward building scheme based on MySQL: the directional crawl and preprocessing multi-source data, design optimization of table structure associated storage, build a three-tier architecture of “database - retrieval module - large model”. Ultimately, a large model in the vertical field of world cultural heritage is formed. Compared with 25 professional questions, the experiment verifies that the big model connected to the database is better in accuracy, completeness and pertinacity, and can accurately call authoritative data. This scheme provides data support for the intelligent protection of cultural heritage. In the future, it will continue to improve from the dimensions of expanding storage, optimizing retrieval, and enriching data to help the protection and inheritance of cultural heritage.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Building an AI Knowledge Infrastructure for the World Cultural Heritage Vertical: Methodology and Challenges from Local Databases

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Yuzhe Liu,
  • Ximeng Wang,
  • Shuai Shao,
  • Lijie Wang

摘要

In order to promote the intelligent protection of world cultural heritage, this paper focuses on the construction of local databases in its vertical AI knowledge infrastructure. Professional dispersion characteristics of cultural heritage data method and general adaptation shortage problem, put forward building scheme based on MySQL: the directional crawl and preprocessing multi-source data, design optimization of table structure associated storage, build a three-tier architecture of “database - retrieval module - large model”. Ultimately, a large model in the vertical field of world cultural heritage is formed. Compared with 25 professional questions, the experiment verifies that the big model connected to the database is better in accuracy, completeness and pertinacity, and can accurately call authoritative data. This scheme provides data support for the intelligent protection of cultural heritage. In the future, it will continue to improve from the dimensions of expanding storage, optimizing retrieval, and enriching data to help the protection and inheritance of cultural heritage.