Filters








9,635 Hits in 6.0 sec

Word Embeddings (Also) Encode Human Personality Stereotypes

Oshin Agarwal, Funda Durupınar, Norman I. Badler, Ani Nenkova
2019 Proceedings of the Eighth Joint Conference on Lexical and Computational Semantics (*  
Here, we present results that show that human stereotypes exist even for much more nuanced judgments such as personality, for a variety of person identities beyond the typically legally protected attributes  ...  Word representations trained on text reproduce human implicit bias related to gender, race and age. Methods have been developed to remove such bias.  ...  We verify that people hold stereotypes about personality and that the human stereotypes can be recovered fairly accurately from word representations.  ... 
doi:10.18653/v1/s19-1023 dblp:conf/starsem/AgarwalDBN19 fatcat:pmzbtzmp3vaczepjnqqa36oqc4

Deconstructing Word Embeddings [article]

Koushik Varma Kalidindi
2019 arXiv   pre-print
A review of Word Embedding Models through a deconstructive approach reveals their several shortcomings and inconsistencies.  ...  A new theoretical embedding model, Derridian Embedding, is proposed in this paper.  ...  Female/male gender stereotypes have appeared on word embeddings trained on Google News data (Bolukbasi et al., 2016) .  ... 
arXiv:1902.00551v1 fatcat:lykupezb3vfxxlngrbl4n7e3y4

Bias in word embeddings

Orestis Papakyriakopoulos, Simon Hegelich, Juan Carlos Medina Serrano, Fabienne Marco
2020 Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency  
Recent studies demonstrate that word embeddings contain and amplify biases present in data, such as stereotypes and prejudice.  ...  Word embeddings are a widely used set of natural language processing techniques that map words to vectors of real numbers.  ...  [4] prove that different meanings of words are 'encoded' in word embeddings and can be retrieved. Zhao et al. [101] propose a methodology to train word embeddings without sexist bias in them.  ... 
doi:10.1145/3351095.3372843 dblp:conf/fat/Papakyriakopoulos20 fatcat:vqus2brtezf7jcamxsm7dcilra

Gender and Racial Stereotype Detection in Legal Opinion Word Embeddings [article]

Sean Matthews, John Hudzina, Dawn Sepehr
2022 arXiv   pre-print
Our analyses using these methods suggest that racial and gender biases are encoded into word embeddings trained on legal opinions.  ...  In this article, we propose an approach for identifying gender and racial stereotypes in word embeddings trained on judicial opinions from U.S. case law.  ...  Bias in Word Embeddings Word embedding approaches such as word2vec (Mikolov et al. 2013a,b) , GloVe (Pennington, Socher, and Manning 2014) , etc., represent words in an n-dimensional space by encoding  ... 
arXiv:2203.13369v2 fatcat:3wix7va6ybbo5kpfa2tnwxcqce

What are the biases in my word embedding? [article]

Nathaniel Swinger, Maria De-Arteaga, Neil Thomas Heffernan IV, Mark DM Leiserson, Adam Tauman Kalai
2019 arXiv   pre-print
We also show how removing names may not remove potential proxy bias.  ...  This paper presents an algorithm for enumerating biases in word embeddings.  ...  with human stereotypes.  ... 
arXiv:1812.08769v4 fatcat:2mqpfww425fi5calsnj2rdyjmq

Human-in-the-Loop Refinement of Word Embeddings [article]

James Powell, Kari Sentz, Martin Klein
2021 arXiv   pre-print
It also allows for better insight into what effect word embeddings, and refinements to word embeddings, have on machine learning pipelines.  ...  Our approach allows a human to identify and address potential quality issues with word embeddings interactively.  ...  Position and distance of word representations within this word embedding space encodes semantic and syntactic information.  ... 
arXiv:2110.02884v1 fatcat:ymdno6bipnevziqitotfp7kwke

Unequal Representations: Analyzing Intersectional Biases in Word Embeddings Using Representational Similarity Analysis [article]

Michael A. Lepori
2020 arXiv   pre-print
We present a new approach for detecting human-like social biases in word embeddings using representational similarity analysis.  ...  Specifically, we probe contextualized and non-contextualized embeddings for evidence of intersectional biases against Black women.  ...  However, detecting these stereotypes is only a proxy for detecting intersectional biases in embeddings: An embedding might encode intersectional differences without encoding more complex interesectional  ... 
arXiv:2011.12086v1 fatcat:qolqbheivrgqdddr2kygloazte

Evaluating Bias In Dutch Word Embeddings [article]

Rodrigo Alejandro Chávez Mulsa, Gerasimos Spanakis
2020 arXiv   pre-print
Recent research in Natural Language Processing has revealed that word embeddings can encode social biases present in the training data which can affect minorities in real world applications.  ...  We implement the Word Embeddings Association Test (WEAT), Clustering and Sentence Embeddings Association Test (SEAT) methods to quantify the gender bias in Dutch word embeddings, then we proceed to reduce  ...  Then we demonstrate that contextualized word embeddings show less bias than traditional ones while also determining that the bias mitigation is not as effective in contextualized word embeddings as in  ... 
arXiv:2011.00244v2 fatcat:ptblujoklrbxta7wywd3wqvr3a

Hurtful Words: Quantifying Biases in Clinical Contextual Word Embeddings [article]

Haoran Zhang, Amy X. Lu, Mohamed Abdalla, Matthew McDermott, Marzyeh Ghassemi
2020 arXiv   pre-print
In this work, we examine the extent to which embeddings may encode marginalized populations differently, and how this may lead to a perpetuation of biases and worsened performance on clinical tasks.  ...  Finally, we explore shortcomings of using adversarial debiasing to obfuscate subgroup information in contextual word embeddings, and recommend best practices for such deep embedding models in clinical  ...  BACKGROUND AND RELATED WORK 2.1 Contextual Embeddings Word embeddings algorithms are methods for numerically representing human text as dense high-dimensional vectors which are amenable to further computational  ... 
arXiv:2003.11515v1 fatcat:jvy2px2s7zejhg64tnxtxoztsq

What are the Biases in My Word Embedding?

Nathaniel Swinger, Maria De-Arteaga, Neil Thomas Heffernan IV, Mark DM Leiserson, Adam Tauman Kalai
2019 Proceedings of the 2019 AAAI/ACM Conference on AI, Ethics, and Society - AIES '19  
We also show how removing names may not remove potential proxy bias.  ...  These embedded biases are concerning in light of the widespread use of word embeddings.  ...  with human stereotypes.  ... 
doi:10.1145/3306618.3314270 dblp:conf/aies/SwingerDHLK19 fatcat:3h6xzu255jgpfoirbbvhlcd5ba

Exploring the Combination of Contextual Word Embeddings and Knowledge Graph Embeddings [article]

Lea Dieudonat, Kelvin Han, Phyllicia Leavitt, Esteban Marquer
2020 arXiv   pre-print
However, their ability to represent the different meanings that a word may have is limited. Such approaches also do not explicitly encode relations between entities, as denoted by words.  ...  "Classical" word embeddings, such as Word2Vec, have been shown to capture the semantics of words based on their distributional properties.  ...  "Static" word embeddings Word embeddings -which may also be called vector-space models or distributional semantics models -are the representation of words as vectors.  ... 
arXiv:2004.08371v1 fatcat:jyj5ce5wifbq5frb63max4rcye

Word embeddings quantify 100 years of gender and ethnic stereotypes

Nikhil Garg, Londa Schiebinger, Dan Jurafsky, James Zou
2018 Proceedings of the National Academy of Sciences of the United States of America  
Word embeddings use vectors to represent words such that the geometry between vectors captures semantic relationship between the words.  ...  In this paper, we develop a framework to demonstrate how the temporal dynamics of the embedding can be leveraged to quantify changes in stereotypes and attitudes toward women and ethnic minorities in the  ...  Comparison with surveys of gender stereotypes. Now, we validate that the historical embeddings also capture gender stereotypes of personality traits.  ... 
doi:10.1073/pnas.1720347115 pmid:29615513 fatcat:sc4qf7ptd5dvjiprm55udegrwa

Word-embeddings Italian semantic spaces: A semantic model for psycholinguistic research

Marco Marelli
2017 Psihologija  
The present paper describes WEISS (Word-Embeddings Italian Semantic Space), a distributional semantic model based on Italian.  ...  The resource is evaluated against two test sets, demonstrating that WEISS obtains a better performance with respect to a baseline encoding word associations.  ...  Remarkably, they were also claimed to represent a more sound model of how humans learn word meanings.  ... 
doi:10.2298/psi161208011m fatcat:s45e35bc3fcftdsywv3yce3znm

Wide range screening of algorithmic bias in word embedding models using large sentiment lexicons reveals underreported bias types [article]

David Rozado
2020 arXiv   pre-print
Interestingly, using the common operationalization of the term bias in the fairness literature, novel types of so far unreported bias types in word embedding models have also been identified.  ...  This work describes a large-scale analysis of sentiment associations in popular word embedding models along the lines of gender and ethnicity but also along the less frequently studied dimensions of socioeconomic  ...  But projecting words onto cultural axes can also reveal that word vectors encode a surprising amount of associations, other than bias, about the empirical world.  ... 
arXiv:1905.11985v6 fatcat:yh5mpr3ecbfb5ces5qlc74jjza

When do Word Embeddings Accurately Reflect Surveys on our Beliefs About People? [article]

Kenneth Joseph, Jonathan H. Morgan
2020 arXiv   pre-print
Social biases are encoded in word embeddings. This presents a unique opportunity to study society historically and at scale, and a unique danger when embeddings are used in downstream applications.  ...  However, we also find that biases in embeddings are much more reflective of survey data for some dimensions of meaning (e.g. gender) than others (e.g. race), and that we can be highly confident that embedding-based  ...  Word embeddings (also) encode human personality stereotypes.  ... 
arXiv:2004.12043v1 fatcat:37bz7vrccbcmlasne6z4xfe3vu
« Previous Showing results 1 — 15 out of 9,635 results