Research

Written work from operating LinkedCulture: a peer-visible paper series documenting what a live multi-institutional retrieval system actually does, and shorter working notes along the way. By Edward Monk.

RSS feed · Papers are archived on Zenodo with DOIs; the pages here are the canonical web editions.

The paper series

  1. Paper No. 3 · June 24, 2026

    Semantic Clusters and Language Partitioning

    Multilingual retrieval is not cross-lingual representation: a 302,798-record embedding space stays 99.9% partitioned by cataloging language, and query translation — not model choice — bridges it.

    Read the web edition · DOI 10.5281/zenodo.20827947

  2. Paper No. 2 · June 10, 2026

    Embedding Cultural Heritage Metadata

    The implementation record: how records from nine institutions become embeddable prose, how OpenSearch and Qdrant retrieval fuse, and where pipeline choices introduce bias.

    Read the web edition · DOI 10.5281/zenodo.20633107

  3. Paper No. 1 · May 27, 2026

    Discovery Architecture for Cultural Heritage

    Keyword search alone cannot bridge the vocabulary gap between catalogers and searchers. The case for layered retrieval that keeps AI at the representation layer, never the interpretation layer.

    Read the web edition · DOI 10.5281/zenodo.20418999

Research notes

Working notes — shorter observations from running the platform — will appear here.