AI & Computingpreprint2026-10-17

Selective gender labeling from ORCID biographies: An open silver standard for evaluating gender-assignment methods

Open access0 citations

Abstract

This preprint presents the ORCID Biography Gender Labeling (OBGL) resource, an openly available, ORCID-linked silver standard designed for the comparative evaluation of gender-assignment methods in scholarly data. OBGL contains 1,016,775 biography-level records in English, Portuguese, and Spanish derived from the 2025 ORCID Public Data File. A transparent lexical procedure assigns 460,057 labels at τ = 1, while the released scores allow users to construct smaller and more selective reference subsets at stricter thresholds. The study examines how marker availability, labeling coverage, and female-to-male label composition vary across languages, affiliation countries, and thresholds. Each released record retains its ORCID iD, enabling linkage to names, affiliations, publication records, and other scholarly metadata required by different gender-assignment approaches. The associated dataset and SQL scripts are available in Zenodo as ORCID Biography Gender Labeling (OBGL), Version 1.1: https://doi.org/10.5281/zenodo.23282271. This manuscript is a preprint and has not yet undergone peer review. Supplementary material: The record also includes Tables S1–S3 accompanying the preprint, with country-by-threshold diagnostics, language-level agreement and abstention results, and the complete language-specific marker vocabularies.

// Source

View paper (DOI)Open access versionOpenAlexZenodo (CERN European Organization for Nuclear Research)Published 2026-10-17

Authors: Pablo Guerrero-Castillo, Jesús M. Álvarez-Llorente, Rodrigo Sánchez Jiménez, Vicente P. Guerrero‐Bote

Institutions: Universidad Complutense de Madrid, Universidad de Extremadura