Digital Collections Explorer

Digital Collections Explorer is an open-source search site for digital collections, with or without a catalog. A gallery, library, archive, or museum points it at a folder, an S3 bucket, or a manifest of images and their catalog. Visitors can then search the collection in plain language ("a steam locomotive", "a handwritten letter with a red seal"), search with a picture, wander through it, and follow every image back to the institution's own record. No catalog or tags are needed, and where a catalog exists, search reads its records too.

The project site shows what it does and the collections searched with it, among them the maps of the Library of Congress from our paper, and holds the documentation. The project started at the Lab for Computing Cultural Heritage at the University of Washington Information School, is developed in the open with contributors from other institutions, and is described in our paper in Computational Humanities Research.

Quick Start

There are three recommended ways to start, each explained in the documentation:

• With an AI coding agent. Ask Claude Code, Codex, Gemini CLI, or Cursor to make your collection searchable. The dce-setup skill in this repository walks it through profiling, indexing, and serving.
• With Docker. One command builds the index, another serves it.
• On a server or AWS. One command keeps a site running. On AWS it shows the cost first.

Citation

If you use Digital Collections Explorer in research, please cite the paper and the release you used, each release having its own DOI on Zenodo.

License

The source code of Digital Collections Explorer is licensed under the MIT License. Documentation and original figures, such as overview.png, are licensed under CC BY 4.0 unless otherwise noted. Third-party code, models, data, and media keep their respective licenses.

Contributing

Contributions are welcome. CONTRIBUTING.md covers setting up for development, running the tests, and sending a change, and the project follows a Code of Conduct.