Understanding Undesirable Word Embedding Associations
Open accessThe DOI matches its Crossref record; 6 of the 6 references it cites were found in a registry or an index.
Abstract
Word embeddings are often criticized for capturing undesirable word associations such as gender stereotypes.However, methods for measuring and removing such biases remain poorly understood.We show that for any embedding model that implicitly does matrix factorization, debiasing vectors post hoc using subspace projection (Bolukbasi et al., 2016) is, under certain conditions, equivalent to training on an unbiased corpus.We also prove that WEAT, the most common association test for word embeddings, systematically overestimates bias.Given that the subspace projection method is provably effective, we use it to derive a new measure of association called the relational inner product association (RIPA).Experiments with RIPA reveal that, on average, skipgram with negative sampling (SGNS) does not make most words any more gendered than they are in the training corpus.However, for gender-stereotyped words, SGNS actually amplifies the gender association in the corpus.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Landauer, T. K., & Dumais, S. T. (1997) · Psychological Review · doi:10.1037/0033-295x.104.2.211 · Cited by 6,282
A unified architecture for natural language processing: deep neural networks with multitask learning
Collobert, R., & Weston, J. (2008) · Proceedings of the 25th international conference on Machine learning - ICML '08 · doi:10.1145/1390156.1390177 · Cited by 5,058 · cited 2×
“in, or vice-versa (Bengio et al., 2003; Mikolov et al., 2013; Collobert and Weston, 2008); (b) low-rank approximations of word-context matrices containing a co-occurrence statistic (Landauer and Dumais, 1997; Levy and Goldberg, 2014).”
BackgroundSemantics derived automatically from language corpora contain human-like biases
Caliskan, A., Bryson, J. J., & Narayanan, A. (2017) · Science · doi:10.1126/science.aal4230 · Cited by 2,887
Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings
Bolukbasi, T., Chang, K.-W., Zou, J., Saligrama, V., & Kalai, A. (2016) · arXiv · doi:10.48550/arxiv.1607.06520 · Cited by 1,615 · cited 17×
“We create lists of biased and appropriate words using the Bolukbasi et al. (2016) lists of gender-biased and gender-appropriate analogies.”
BackgroundMethodResultInfluentialGonen, H., & Goldberg, Y. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1061 · Cited by 454 · cited 1×
“This may explain why it is still possible to cluster words by gender after debiasing them using a handful of gender-defining word pairs (Gonen and Goldberg, 2019).”
BackgroundA Latent Variable Model Approach to PMI-based Word Embeddings
Arora, S., Li, Y., Liang, Y., Ma, T., & Risteski, A. (2016) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00106 · Cited by 286 · cited 1×
“(Arora et al., 2016; Ethayarajh et al., 2018).”
Cited by
On the Opportunities and Risks of Foundation Models
Bommasani, R., Hudson, D. A., Adeli, E., Altman, R., Arora, S., von Arx, S., Bernstein, M. S., Bohg, J., Bosselut, A., Brunskill, E., et al. (2021) · arXiv · doi:10.48550/arxiv.2108.07258 · Cited by 2,676
Self-citationNull It Out: Guarding Protected Attributes by Iterative Nullspace Projection
Ravfogel, S., Elazar, Y., Gonen, H., Twiton, M., & Goldberg, Y. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.647 · Cited by 287
Utility is in the Eye of the User: A Critique of NLP Leaderboards
Ethayarajh, K., & Jurafsky, D. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.393 · Cited by 147
Self-citationTowards trustworthy LLMs: a review on debiasing and dehallucinating in large language models
Lin, Z., Guan, S., Zhang, W., Zhang, H., Li, Y., & Zhang, H. (2024) · Artificial Intelligence Review · doi:10.1007/s10462-024-10896-y · Cited by 141
Barikeri, S., Lauscher, A., Vulić, I., & Glavaš, G. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.151 · Cited by 121
“, 2019), analyses of the proposed measures (e.g., Gonen and Goldberg, 2019; Ethayarajh et al., 2019), more comprehensive evaluation frameworks (Lauscher et al.”
Background
Metadata
- Authors
- Ethayarajh, K., Duvenaud, D., & Hirst, G.
- Venue
- Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics
- Published
- 2019
- Pages
- 1696-1705
- OpenAlex
- W2952349219
- Cited by
- 109
- References
- 6
- FWCI
- 19.92
- Type
- conference paper
- Language
- en
- Open access
- hybrid
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.