On Measuring Social Biases in Sentence Encoders
Explore this paper's citation graph
Summary
The Word Embedding Association Test is extended to measure bias in sentence encoders and mixed results including suspicious patterns of sensitivity that suggest the test’s assumptions may not hold in general.
- Type
- preprint
- Published
- 2019-03-25
- Cited by
- 742
- References
- 51
- Access
- Open access
- OpenAlex
- https://openalex.org/W2922862135
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:85518027
Keywords
Sentence, Word (group theory), Computer science, Encoder, Test (biology)
References
- Sister Citizen: Shame, Stereotypes, and Black Women in America
- Black Sexual Politics: African Americans, Gender, and the New Racism
- Aligning Books and Movies: Towards Story-Like Visual Explanations by Watching Movies and Reading Books
- A large annotated corpus for learning natural language inference
- Fast-Track Women and the “Choice” to Stay Home
- Raunch versus prude: contemporary sex blogs and erotic memoirs by women
- sister citizen: shame, stereotypes, and black women in America
- Crazy Patriotism and Angry (Post)Black Women
- Black sexual politics: African Americans, gender, and the new racism
- Understanding and using the implicit association test: I. An improved scoring algorithm.
- Discrimination in online ad delivery
- The measurement of psychological androgyny.
- Understanding and using the Implicit Association Test: III. Meta-analysis of predictive validity.
- A Simple Sequentially Rejective Multiple Test Procedure
- Penalties for success: reactions to women who succeed at male gender-typed tasks.
- Measuring individual differences in implicit cognition: the implicit association test.
- Distributed Representations of Words and Phrases and their Compositionality
- Deep Unordered Composition Rivals Syntactic Methods for Text Classification
- GloVe: Global Vectors for Word Representation
- Distributional techniques for philosophical enquiry
Cited by
- Bias in semantic and discourse interpretation
- Gender Bias in Neural Natural Language Processing
- Gender Bias in Contextualized Word Embeddings
- Fair Is Better than Sensational: Man Is to Doctor as Woman Is to Doctor
- Conceptor Debiasing of Word Representations Evaluated on WEAT
- Mitigating Gender Bias in Natural Language Processing: Literature Review
- Good Secretaries, Bad Truck Drivers? Occupational Gender Stereotypes in Sentiment Analysis
- A Survey on Bias and Fairness in Machine Learning
- Examining Gender Bias in Languages with Grammatical Gender
- Assessing Social and Intersectional Biases in Contextualized Word Representations
- Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks
- Pretrained AI Models: Performativity, Mobility, and Change
- Measuring Bias in Contextualized Word Representations
- Transfer Learning from Pre-trained BERT for Pronoun Resolution
- Perturbation Sensitivity Analysis to Detect Unintended Model Biases
- The Role of Protected Class Word Lists in Bias Identification of Contextualized Word Representations
- Gendered Ambiguous Pronoun (GAP) Shared Task at the Gender Bias in NLP Workshop 2019
- What do Deep Networks Like to Read?
- Unintended machine learning biases as social barriers for persons with disabilitiess
- Man is to Person as Woman is to Location: Measuring Gender Bias in Named Entity Recognition
Related papers
- GloVe: Global Vectors for Word Representation
- BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
- Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings
- Semantics derived automatically from language corpora contain human-like biases
- Lipstick on a Pig: Debiasing Methods Cover up Systematic Gender Biases in Word Embeddings But do not Remove Them
- Learning Gender-Neutral Word Embeddings
- Measuring and Mitigating Unintended Bias in Text Classification
- Distributed Representations of Words and Phrases and their Compositionality