A General Framework for Implicit and Explicit Debiasing of Distributional Word Vector Spaces

  title={A General Framework for Implicit and Explicit Debiasing of Distributional Word Vector Spaces},
  author={Anne Lauscher and Goran Glavas and Simone Paolo Ponzetto and Ivan Vulic},
  • Anne Lauscher, Goran Glavas, +1 author Ivan Vulic
  • Published 2020
  • Computer Science
  • ArXiv
  • Distributional word vectors have recently been shown to encode many of the human biases, most notably gender and racial biases, and models for attenuating such biases have consequently been proposed. [...] Key Method We then propose three debiasing models that operate on explicit or implicit bias specifications, and that can be composed towards more robust debiasing. Finally, we devise a full-fledged evaluation framework in which we couple existing bias metrics with newly proposed ones. Experimental findings…Expand Abstract
    9 Citations

    Figures, Tables, and Topics from this paper.

    OSCaR: Orthogonal Subspace Correction and Rectification of Biases in Word Embeddings
    • 2
    • Highly Influenced
    • PDF
    Predictive Biases in Natural Language Processing Models: A Conceptual Framework and Overview
    • 17
    • PDF
    Measuring Societal Biases in Text Corpora via First-Order Co-occurrence
    • 1
    • PDF
    AraWEAT: Multidimensional Analysis of Biases in Arabic Word Embeddings
    The Geometry of Distributed Representations for Better Alignment, Attenuated Bias, and Improved Interpretability


    Are We Consistently Biased? Multidimensional Analysis of Biases in Distributional Word Vectors
    • 7
    • PDF
    Lipstick on a Pig: Debiasing Methods Cover up Systematic Gender Biases in Word Embeddings But do not Remove Them
    • 145
    • Highly Influential
    • PDF
    Attenuating Bias in Word Vectors
    • 31
    • Highly Influential
    • PDF
    Gender Bias in Contextualized Word Embeddings
    • 94
    • Highly Influential
    • PDF
    Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings
    • 884
    • Highly Influential
    • PDF
    Learning Gender-Neutral Word Embeddings
    • 116
    • Highly Influential
    • PDF
    Black is to Criminal as Caucasian is to Police: Detecting and Removing Multiclass Bias in Word Embeddings
    • 51
    • Highly Influential
    • PDF
    Reducing Gender Bias in Abusive Language Detection
    • 77
    • PDF
    Separated by an Un-common Language: Towards Judgment Language Informed Vector Space Modeling
    • 52
    • PDF