English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

We study the effect of tokenization on gender bias in machine translation, anaspect that has been largely overlooked in previous works. Specifically, wefocus on the interactions between the frequency of gendered profession names intraining data, their representation in the subword tokenizer's vocabulary, andgender bias. We observe that female and non-stereotypical gender inflections ofprofession names (e.g., Spanish "doctora" for "female doctor") tend to be splitinto multiple subword tokens. Our results indicate that the imbalance of genderforms in the model's training corpus is a major factor contributing to genderbias and has a greater impact than subword splitting. We show that analyzingsubword splits provides good estimates of gender-form imbalance in the trainingdata and can be used even when the corpus is not publicly available. We alsodemonstrate that fine-tuning just the token embedding layer can decrease thegap in gender prediction accuracy between female and male forms withoutimpairing the translation quality.

Exploring the Impact of Training Data Distribution and Subword Tokenization on Gender Bias in Machine Translation