The Shape of Biological Metadata: Measuring Repository Richness with Entity-Based NLP Metrics
- 1Centre for Engineering Biology, School of Biological Sciences, University of Edinburgh, Edinburgh EH9 3BF, UK
Abstract
Ensuring the availability and accessibility of research data is fundamental to advancing knowledge, as codified in the FAIR principles (Findable, Accessible, Interoperable, and Reusable). Accurate metadata documentation is indispensable for meeting these principles; however, entries in deposition databases often contain inadequate, repetitive, or incomplete descriptions. Much of this metadata is captured in free-text fields, motivating the need for scalable, repository-agnostic methods to quantify metadata richness. Here, we quantify free-text metadata richness across three repositories using Natural Language Processing (NLP) methods: BioDare2, an experimental circadian rhythm database; DataShare, a domain-agnostic University of Edinburgh database; and Image Data Resource (IDR), a public repository of biological image datasets from published studies. In general, repositories exhibit distinct distributions of word count and information density, consistent with differences in scope. Named-entity recognition and information-density metrics detected significant category-level differences in BioDare2 (species) and DataShare (communities), while identifying greater consistency in the more curated IDR. The most frequent entities reflected each repository's focus: circadian terminology in BioDare2, microscopy-related entities in IDR, and community-driven terms in DataShare. We develop a scalable framework utilising word counts, named-entity recognition, and entity-derived information density to assess metadata quality across repositories, offering a broadly applicable evaluation tool.
Related
- DBCLS BioHackathon 2026 report: An interactive, harmonised BioSample metadata resource for biological sample discovery
- From APTs to advanced persistent biological threats (APBTs): a taxonomy and large-scale metadata audit of genomic surveillance infrastructure
- CADS: Measuring the Incremental Contribution of LLM Proposers in Sequential Biological Screens
- A sampling-theory framework for richness-uniqueness relationships
- Decoupling of above- and below-ground intraspecific trait variation along richness gradients in plants
BibTeX
@misc{rodriguezcubillos2026shape,
title = {{The Shape of Biological Metadata: Measuring Repository Richness with Entity-Based NLP Metrics}},
author = {M. J. Rodriguez-Cubillos and T. Zielinski and J. R. Swedlow and T. I. Simpson and A. J. Millar},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.08.749906},
url = {https://www.biorxiv.org/content/10.64898/2026.09.08.749906v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - The Shape of Biological Metadata: Measuring Repository Richness with Entity-Based NLP Metrics
AU - M. J. Rodriguez-Cubillos
AU - T. Zielinski
AU - J. R. Swedlow
AU - T. I. Simpson
AU - A. J. Millar
PY - 2026
DA - 2026/09/08
PB - bioRxiv
DO - 10.64898/2026.09.08.749906
UR - https://www.biorxiv.org/content/10.64898/2026.09.08.749906v1
AB - Ensuring the availability and accessibility of research data is fundamental to advancing knowledge, as codified in the FAIR principles (Findable, Accessible, Interoperable, and Reusable). Accurate metadata documentation is indispensable for meeting these principles; however, entries in deposition databases often contain inadequate, repetitive, or incomplete descriptions. Much of this metadata is captured in free-text fields, motivating the need for scalable, repository-agnostic methods to quantify metadata richness. Here, we quantify free-text metadata richness across three repositories using Natural Language Processing (NLP) methods: BioDare2, an experimental circadian rhythm database; DataShare, a domain-agnostic University of Edinburgh database; and Image Data Resource (IDR), a public repository of biological image datasets from published studies. In general, repositories exhibit distinct distributions of word count and information density, consistent with differences in scope. Named-entity recognition and information-density metrics detected significant category-level differences in BioDare2 (species) and DataShare (communities), while identifying greater consistency in the more curated IDR. The most frequent entities reflected each repository's focus: circadian terminology in BioDare2, microscopy-related entities in IDR, and community-driven terms in DataShare. We develop a scalable framework utilising word counts, named-entity recognition, and entity-derived information density to assess metadata quality across repositories, offering a broadly applicable evaluation tool.
ER -