Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Garg, Sarthak, Gheini, Mozhdeh, Emmanuel, Clara, Likhomanenko, Tatiana, Gao, Qin, Paulik, Matthias
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2407.20438
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911971486466048
author Garg, Sarthak
Gheini, Mozhdeh
Emmanuel, Clara
Likhomanenko, Tatiana
Gao, Qin
Paulik, Matthias
author_facet Garg, Sarthak
Gheini, Mozhdeh
Emmanuel, Clara
Likhomanenko, Tatiana
Gao, Qin
Paulik, Matthias
contents Machine translation (MT) systems often translate terms with ambiguous gender (e.g., English term "the nurse") into the gendered form that is most prevalent in the systems' training data (e.g., "enfermera", the Spanish term for a female nurse). This often reflects and perpetuates harmful stereotypes present in society. With MT user interfaces in mind that allow for resolving gender ambiguity in a frictionless manner, we study the problem of generating all grammatically correct gendered translation alternatives. We open source train and test datasets for five language pairs and establish benchmarks for this task. Our key technical contribution is a novel semi-supervised solution for generating alternatives that integrates seamlessly with standard MT models and maintains high performance without requiring additional components or increasing inference overhead.
format Preprint
id arxiv_https___arxiv_org_abs_2407_20438
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Generating Gender Alternatives in Machine Translation
Garg, Sarthak
Gheini, Mozhdeh
Emmanuel, Clara
Likhomanenko, Tatiana
Gao, Qin
Paulik, Matthias
Computation and Language
Artificial Intelligence
Machine translation (MT) systems often translate terms with ambiguous gender (e.g., English term "the nurse") into the gendered form that is most prevalent in the systems' training data (e.g., "enfermera", the Spanish term for a female nurse). This often reflects and perpetuates harmful stereotypes present in society. With MT user interfaces in mind that allow for resolving gender ambiguity in a frictionless manner, we study the problem of generating all grammatically correct gendered translation alternatives. We open source train and test datasets for five language pairs and establish benchmarks for this task. Our key technical contribution is a novel semi-supervised solution for generating alternatives that integrates seamlessly with standard MT models and maintains high performance without requiring additional components or increasing inference overhead.
title Generating Gender Alternatives in Machine Translation
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2407.20438