Evaluating the Underlying Gender Bias in Contextualized Word Embeddings

Abstract

Gender bias is highly impacting natural language processing applications. Word embeddings have clearly been proven both to keep and amplify gender biases that are present in current data sources. Recently, contextualized word embeddings have enhanced previous word embedding techniques by computing word vector representations dependent on the sentence they appear in. In this paper, we study the impact of this conceptual change in the word embedding computation in relation with gender bias. Our analysis includes different measures previously applied in the literature to standard word embeddings. Our findings suggest that contextualized word embeddings are less biased than standard ones even when the latter are debiased.

References in corpus (5)

• Distributed Representations of Words and Phrases and their Compositionality
• UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction
• Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings
• Gender Bias in Coreference Resolution: Evaluation and Debiasing Methods
• Learning Gender-Neutral Word Embeddings

Cited by in corpus (7)

• Unmasking Contextual Stereotypes: Measuring and Mitigating BERT's Gender Bias
• Hurtful Words: Quantifying Biases in Clinical Contextual Word Embeddings
• Measuring Bias in Contextualized Word Representations
• GeBioToolkit: Automatic Extraction of Gender-Balanced Multilingual Corpus of Wikipedia Biographies
• Intersectional Bias in Causal Language Models
• Gender Bias in Multilingual Neural Machine Translation: The Architecture Matters
• Examining Gender Bias in Languages with Grammatical Gender