Unsupervised Discovery of Implicit Gender Bias

Field, Anjalie; Tsvetkov, Yulia

doi:10.18653/v1/2020.emnlp-main.44

Cited by 31 publications

(28 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…But veiled offenses are not represented in existing toxicity datasets (Waseem and Hovy, 2016;Davidson et al, 2017;Founta et al, 2018) and building a new dataset is expensive: candidates for annotation cannot be filtered through lexicons and random sampling of social media posts will surface only a tiny fraction of relevant examples (Breitfeller et al, 2019). Moreover, since biased text is often unconscious and subjective, untrained annotators might mislabel it due to their own biases (Breitfeller et al, 2019;Field and Tsvetkov, 2020).…”

Section: Toxic Language In Disguisementioning

confidence: 99%

Fortifying Toxic Speech Detectors Against Veiled Toxicity

Han¹,

Tsvetkov²

2020

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)

Self Cite

View full text Add to dashboard Cite

Modern toxic speech detectors are incompetent in recognizing disguised offensive language, such as adversarial attacks that deliberately avoid known toxic lexicons, or manifestations of implicit bias. Building a large annotated dataset for such veiled toxicity can be very expensive. In this work, we propose a framework aimed at fortifying existing toxic speech detectors without a large labeled corpus of veiled toxicity. Just a handful of probing examples are used to surface orders of magnitude more disguised offenses. We augment the toxic speech detector's training data with these discovered offensive examples, thereby making it more robust to veiled toxicity while preserving its utility in detecting overt toxicity. 1 Warning: this paper contains examples that may be offensive or upsetting.

show abstract

Section: Toxic Language In Disguisementioning

confidence: 99%

Fortifying Toxic Speech Detectors Against Veiled Toxicity

Han¹,

Tsvetkov²

2020

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)

Self Cite

View full text Add to dashboard Cite

show abstract

“…One straightforward reason is that implicit stereotypes persist in our society. Even if explicit gender discrimination occurs less frequently today than in the past, implicit attitudes about females being submissive and less worthy than males remain pervasive (Field & Tsvetkov, 2020;Storage et al, 2020). Consistent with this possibility is the observation that males are considered more prototypical than females when categorizing humans (Bailey et al, 2020).…”

Section: Patterns Of Gender Representation Explainedmentioning

confidence: 97%

Sixty Years of Gender Representation in Children’s Books: Conditions Associated with Overrepresentation of Male versus Female Protagonists

Casey¹,

Novick²,

Sf³

2020

Preprint

View full text Add to dashboard Cite

As a reflection of prominent cultural norms, children’s literature plays an integral role in the acquisition and development of societal attitudes. Previous reports of male overrepresentation in books targeted towards children are consistent with a history of gender disparity across media and society. However, it is unknown whether such bias has been attenuated in recent years with increasing emphasis on gender equity and greater accessibility of books. Here, we provide an up-to-date estimate of the relative proportion of male and female protagonists in 3,288 children’s books (0-16 years) published between 1960-2020. We find that although the proportion of female protagonists has increased over this 60-year period, male protagonists remain overrepresented even in recent years. Importantly, we also find stable effects related to author gender, age of the target audience, book genre (fiction vs. non-fiction), and character type (human vs. non-human) on the male-to-female ratio of protagonists. We suggest that this comprehensive account of the factors influencing the rates of appearance of male and female protagonists can be leveraged to develop specific recommendations for promoting more equitable gender representation in children’s literature, with important consequences for child development and society.

show abstract

“…Thus, in many existing datasets implicit abuse can be found in only a small proportion of instances (Wiegand et al, 2019). Furthermore, implicit abuse presents additional challenges to human annotators as sometimes specific background knowledge and experience are required in order to understand the hidden meaning behind implicit statements (Sap et al, 2020;Breitfeller et al, 2019;Field & Tsvetkov, 2020). To deal effectively with this class of abuse, annotated datasets focusing on implicitly abusive language are needed so that automatic detection systems are exposed to a wide variety of such examples through their training data (Wiegand et al, 2021).…”

Section: Professional Responsibilitymentioning

confidence: 99%

Confronting Abusive Language Online: A Survey from the Ethical and Human Rights Perspective

Kiritchenko

Nejadgholi

Fraser

2021

jair

View full text Add to dashboard Cite

The pervasiveness of abusive content on the internet can lead to severe psychological and physical harm. Significant effort in Natural Language Processing (NLP) research has been devoted to addressing this problem through abusive content detection and related sub-areas, such as the detection of hate speech, toxicity, cyberbullying, etc. Although current technologies achieve high classification performance in research studies, it has been observed that the real-life application of this technology can cause unintended harms, such as the silencing of under-represented groups. We review a large body of NLP research on automatic abuse detection with a new focus on ethical challenges, organized around eight established ethical principles: privacy, accountability, safety and security, transparency and explainability, fairness and non-discrimination, human control of technology, professional responsibility, and promotion of human values. In many cases, these principles relate not only to situational ethical codes, which may be context-dependent, but are in fact connected to universal human rights, such as the right to privacy, freedom from discrimination, and freedom of expression. We highlight the need to examine the broad social impacts of this technology, and to bring ethical and human rights considerations to every stage of the application life-cycle, from task formulation and dataset design, to model training and evaluation, to application deployment. Guided by these principles, we identify several opportunities for rights-respecting, socio-technical solutions to detect and confront online abuse, including ‘nudging’, ‘quarantining’, value sensitive design, counter-narratives, style transfer, and AI-driven public education applications.evaluation, to application deployment. Guided by these principles, we identify several opportunities for rights-respecting, socio-technical solutions to detect and confront online abuse, including 'nudging', 'quarantining', value sensitive design, counter-narratives, style transfer, and AI-driven public education applications.

show abstract

Unsupervised Discovery of Implicit Gender Bias

Cited by 31 publications

References 38 publications

Fortifying Toxic Speech Detectors Against Veiled Toxicity

Fortifying Toxic Speech Detectors Against Veiled Toxicity

Sixty Years of Gender Representation in Children’s Books: Conditions Associated with Overrepresentation of Male versus Female Protagonists

Confronting Abusive Language Online: A Survey from the Ethical and Human Rights Perspective

Contact Info

Product

Resources

About