[Paper Review] Mitigating Gender Bias in Natural Language Processing: Literature Review

Tony Sun, Andrew Gaut+8 more|arXiv (Cornell University)|Jun 21, 2019

Hate Speech and Cyberbullying Detection56 references41 citations

TL;DR

This paper surveys recognizing and mitigating gender bias in NLP, catalogs four forms of representation bias, reviews debiasing methods, and discusses their advantages, drawbacks, and future research directions.

ABSTRACT

As Natural Language Processing (NLP) and Machine Learning (ML) tools rise in popularity, it becomes increasingly vital to recognize the role they play in shaping societal biases and stereotypes. Although NLP models have shown success in modeling various applications, they propagate and may even amplify gender bias found in text corpora. While the study of bias in artificial intelligence is not new, methods to mitigate gender bias in NLP are relatively nascent. In this paper, we review contemporary studies on recognizing and mitigating gender bias in NLP. We discuss gender bias based on four forms of representation bias and analyze methods recognizing gender bias. Furthermore, we discuss the advantages and drawbacks of existing gender debiasing methods. Finally, we discuss future studies for recognizing and mitigating gender bias in NLP.

Motivation & Objective

• Motivate the study of gender bias in NLP and its societal impact.
• Identify and categorize forms of representation bias in NLP.
• Survey methods for recognizing gender bias in NLP systems.
• Evaluate advantages and drawbacks of existing gender debiasing techniques.
• Suggest directions for future research in recognizing and mitigating gender bias in NLP.

Proposed method

• Conduct a literature review of contemporary NLP bias studies.
• Categorize representation bias into four forms.
• Analyze recognition and mitigation methods for gender bias.
• Discuss advantages and drawbacks of debiasing approaches.
• Propose directions for future research in bias recognition and mitigation.

Experimental results

Research questions

• RQ1 What are the four forms of representation bias in NLP?
• RQ2 What methods exist to recognize gender bias in NLP models and outputs?
• RQ3 What are the advantages and drawbacks of existing gender debiasing techniques?
• RQ4 What future research directions are promising for recognizing and mitigating gender bias in NLP?

Key findings

• Identifies and outlines four forms of representation bias in NLP.
• Reviews methods for recognizing gender bias in NLP systems.
• Analyzes the pros and cons of current gender debiasing approaches.
• Discusses limitations and potential directions for future research in bias mitigation.