Are We Consistently Biased? Multidimensional Analysis of Biases in Distributional Word Vectors

By Anne Lauscher, Goran Glavas. 2019

58 citations recorded by the source corpus.

Word embeddings have recently been shown to reflect many of the pronounced societal biases (e.g., gender bias or racial bias). Existing studies are, however, limited in scope and do not investigate the consistency of biases across relevant dimensions like embedding models, types of texts, and different languages. In this work, we present a systematic study of biases encoded in distributional word vector spaces: we analyze how consistent the bias effects are across languages, corpora, and embedding models. Furthermore, we analyze the cross-lingual biases encoded in bilingual embedding spaces, indicative of the effects of bias transfer encompassed in cross-lingual transfer of NLP models. Our study yields some unexpected findings, e.g., that biases can be emphasized or downplayed by different embedding models or that user-generated content may be less biased than encyclopedic text. We hope our work catalyzes bias research in NLP and informs the development of bias reduction techniques.

Subfields: Artificial Intelligence, Representation & Self-Supervised Learning, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Representation & Self-Supervised Learning

In Safety, Robustness & Interpretability

Citations in the corpus

Language (Technology) is Power: A Critical Survey of “Bias” in NLP (2020, background): “…(2016); Larson (2017); McCurdy and Serbetçi (2017); Schnoebelen (2017); Basta et al. (2019); Ethayarajh et al. (2019); Gonen and Goldberg (2019); Lauscher and Glavaš (2019); Loukina et al. (2019); Mayfield et al. (2019); Mirzaev et al. (2019); Prabhumoye et al. (2019); Ruane et al. (2019);…”
Towards Understanding and Mitigating Social Biases in Language Models (2021, background): “For example, word embeddings are shown to reflect and propagate social biases in the form of undesirable associations that reinforce negative stereotypes about particular social groups (Lauscher and Glavaš, 2019; Caliskan et al., 2017; Bolukbasi et al., 2016).”
Towards Debiasing Sentence Representations (2020, background): “Although word embeddings [19, 16] are useful for a variety of tasks, recent work has shown that they reflect and propagate social biases present in training corpora [12, 6, 25, 5].”
A Survey of Race, Racism, and Anti-Racism in NLP (2021, background): “While identity terms can be examined across languages (van Miltenburg et al., 2017), differences in naming conventions often do not translate, leading some studies to omit examining racial bias in non-English languages (Lauscher and Glavaˇs, 2019).”
JW300: A Wide-Coverage Parallel Corpus for Low-Resource Languages (2019, background): “, 2017; Dev and Phillips, 2019; Gonen and Goldberg, 2019), especially in multilingual settings (Lauscher and Glavaš, 2019).”
Theories of “Gender” in NLP Bias Research (2022, unknown): “07337 [81] Anne Lauscher and Goran Glavaš.”
AraWEAT: Multidimensional Analysis of Biases in Arabic Word Embeddings (2020, methodology): “As a first step in the analysis of language technology biases for Arabic, we present ARAWEAT, an Arabic extension to the multilingual XWEAT framework (Lauscher and Glavaš, 2019).”
Gender Bias in Multilingual Embeddings and Cross-Lingual Transfer (2020, background): “…in various components of the NLP systems, such as the training dataset (Zhao et al., 2018a; Rudinger et al., 2018), the embed-dings (Bolukbasi et al., 2016; Caliskan et al., 2017; Zhou et al., 2019; Manzini et al., 2019) as well as the pre-trained models (Zhao et al., 2019; Kurita et al., 2019).”
Pipelines for Social Bias Testing of Large Language Models (2022, background): “…studies have been conducted to analyse and determine the level of bias in static word embed-dings in binary and multi-class scenarios (Boluk-basi et al., 2016; Caliskan et al., 2017; Garg et al., 2018; Swinger et al., 2019; Manzini et al., 2019; Lauscher and Glavaš, 2019; Gonen and Goldberg, 2019).”
Intrinsic Bias Metrics Do Not Correlate with Application Bias (2020, methodology): “Lauscher and Glavas (2019) suggest that this difference may cause bias to be acquired and encoded differently in fastText and word2vec (We discuss this in more detail in Section 5).”
A General Framework for Implicit and Explicit Debiasing of Distributional Word Vector Spaces (2019, methodology): “Such biases are observed across languages and embedding methods (Lauscher and Glavaˇs 2019), both in static and contextualized word embeddings (Zhao et al. 2019).”