Identifying Gender Stereotypes and Biases in Automated Translation from English to Italian using Similarity Networks

2025-02-17

AI Summary

This study addresses gender bias in English-to-Italian machine translation (MT), aiming to enhance gender inclusivity and objectivity. Methodologically, it formalizes the legally and linguistically grounded principle of “gender neutrality”—as defined by international human rights law and linguistics—into a computable metric. A semantic network analysis framework is developed, leveraging cosine similarity between word embeddings to quantify the magnitude and direction of bias in gendered pronoun translations (e.g., *she/he* → *lei/lui*) relative to contextual words. Its key contribution lies in a novel interdisciplinary bias modeling approach integrating law, linguistics, and AI, yielding an interpretable and reproducible bias measurement. Empirical evaluation systematically diagnoses gender bias across state-of-the-art MT systems, providing evidence-based insights and actionable intervention strategies for training gender-neutral translation models.

Technology Category

• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Machine Learning: Ethics, Bias, and Fairness
• Philosophy and Ethics of AI: Bias, Fairness & Equity

Application Category

• Social Networks and Social Media: Fairness and bias in social network and social media analysis
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

This paper is a collaborative effort between Linguistics, Law, and Computer Science to evaluate stereotypes and biases in automated translation systems. We advocate gender-neutral translation as a means to promote gender inclusion and improve the objectivity of machine translation. Our approach focuses on identifying gender bias in English-to-Italian translations. First, we define gender bias following human rights law and linguistics literature. Then we proceed by identifying gender-specific terms such as she/lei and he/lui as key elements. We then evaluate the cosine similarity between these target terms and others in the dataset to reveal the model's perception of semantic relations. Using numerical features, we effectively evaluate the intensity and direction of the bias. Our findings provide tangible insights for developing and training gender-neutral translation algorithms.

Problem

• Identify gender stereotypes in translations
• Evaluate biases in English-to-Italian machine translation
• Promote gender-neutral algorithms for inclusion

Innovation

• gender-neutral translation algorithms
• cosine similarity evaluation
• gender-specific terms identification