What the study found
Generative artificial intelligence (AI) was used to improve descriptive metadata for digitized photograph collections in an academic library. The study found that this approach produced substantial improvements over existing minimal metadata.
Why the authors say this matters
The authors say this matters because better descriptive metadata can enhance discoverability and accessibility for digitized photograph collections. They also conclude that the case study offers practical guidance for cultural heritage institutions considering AI in their workflows.
What the researchers tested
The researchers tested vision-enabled and multimodal large language models for generating Metadata Object Description Schema-compliant metadata from digitized photographic prints. After a pilot test, Anthropic’s Claude Sonnet 4 was selected for production use, and a Python-based process was used to fit the work into existing digital collections workflows.
What worked and what didn't
The production process created descriptive metadata for 2,263 digitized photos. Generated subject terms mapped successfully to the Faceted Application of Subject Terminology vocabulary in 64% of cases, and the project team also added transparency notices about AI involvement in metadata creation.
What to keep in mind
The abstract does not describe detailed limitations. It does note that the case study was conducted in an academic library and was intended as an example that could be adapted to different institutional contexts and collection sizes.
Key points
- AI was used to improve descriptive metadata for digitized photograph collections in an academic library.
- The study found substantial improvement over existing minimal descriptive metadata.
- Claude Sonnet 4 was selected after a pilot test and used to create metadata for 2,263 digitized photos.
- Subject terms matched the Faceted Application of Subject Terminology vocabulary in 64% of cases.
- The project added transparency notices about AI involvement in metadata creation.
Disclosure
- Research title:
- AI improved descriptive metadata for digitized photographs
- Authors:
- Erin Wolfe
- Institutions:
- University of Kansas
- Publication date:
- 2026-02-23
- OpenAlex record:
- View
Get the weekly research newsletter
Stay current with peer-reviewed research without reading academic papers — one filtered digest, every Friday.