Document is current

Article History

Received: 7 January 2025

Accepted: 3 June 2025

First Online: 27 June 2025

Declarations

:

: The author declares no competing interest.

: In this study, gender labels were not disclosed by users but were heuristically inferred from anonymized features such as profile descriptions and timezone metadata. No personally identifiable information (PII) was collected or used, and all data processing adhered to ethical research standards under institutional review board (IRB) approval. Inferring gender may introduce ethical concerns, including misclassifications, reinforcement of binary gender norms, and demographic bias, as emphasized in recent literature [–]. Although the proposed model utilizes regularization and class balancing to improve fairness, these techniques may not fully mitigate latent bias. Due to ethical and institutional constraints, the dataset and label inference methods cannot be shared.

: Not applicable.