Corpus of Multilingual Gender-Specified Fashion Prompts

This dataset provides a multilingual corpus of fashion-related prompts designed for the analysis of gender bias in text-to-image generation systems. It includes Spanish and Chinese prompts, with and without nationality specification. For each language–nationality configuration, the dataset contains three controlled variants (male, female, and neutral) derived from the same base prompts.

Each CSV file corresponds to one language–nationality setting and contains 100 base prompts expressed as three gender-specific textual variants. The dataset is intended for reproducible evaluation of gender representation and cross-linguistic effects in multimodal generative models, particularly within the fashion domain.