Skip to main content

Data from: Wide range screening of algorithmic bias in word embedding models using large sentiment lexicons reveals underreported bias types

Cite this dataset

Rozado, David (2020). Data from: Wide range screening of algorithmic bias in word embedding models using large sentiment lexicons reveals underreported bias types [Dataset]. Dryad.


Concerns about gender bias in word embedding models have captured substantial attention in the algorithmic bias research literature. Other bias types however have received lesser amounts of scrutiny. 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 status, age, physical appearance, sexual orientation, religious sentiment and political leanings. Consistent with previous scholarly literature, this work has found systemic bias against given names popular among African-Americans in most embedding models examined. Gender bias in embedding models however appears to be multifaceted and often reversed in polarity to what has been regularly reported. 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. Specifically, the popular embedding models analyzed here display negative biases against middle and working-class socioeconomic status, male children, senior citizens, plain physical appearance and intellectual phenomena such as Islamic religious faith, non-religiosity and conservative political orientation. Reasons for the paradoxical underreporting of these bias types in the relevant literature are probably manifold but widely held blind spots when searching for algorithmic bias and a lack of widespread technical jargon to unambiguously describe a variety of algorithmic associations could conceivably be playing a role. The causal origins for the multiplicity of loaded associations attached to distinct demographic groups within embedding models are often unclear but the heterogeneity of said associations and their potential multifactorial roots raises doubts about the validity of grouping them all under the umbrella term bias. Richer and more fine-grained terminology as well as a more comprehensive exploration of the bias landscape could help the fairness epistemic community to characterize and neutralize algorithmic discrimination more efficiently.


This data set has collected several popular pre-trained word embedding models. 

-Word2vec Skip-Gram trained on Google News corpus (100B tokens)

-Glove trained on Wikipedia 2014 + Gigaword 5 (6B tokens)

-Glove trained on 2B tweets Twitter corpus (27B tokens)

-Glove trained on Common Crawl (42B tokens)

-Glove trained on Common Crawl (840B tokens)

-FastText trained with subword infomation on Wikipedia 2017, UMBC webbase corpus and news dataset (16B tokens)

-Fastext trained with subword infomation on Common Crawl (600B tokens)"

Usage notes

The python library gensim ( contains several utility functions to read this data set

The file "fastext-common-crawl-300d-2M-subword-gensim-keyedvectors-format.rar" contains the fasttext model in word2vec format, which can be loaded with gensim. Following is a code snippet for loading the data:

from gensim.models import KeyedVectors
modelsPath = os.path.normpath(r'PATH TO FILE')
modelFilePath = os.path.join(modelsPath,
model = KeyedVectors.load(modelFilePath)