2020 · 267 citations · 26 references
ComputingComputer ScienceComputational LinguisticsDisabilityNatural Language ProcessingDisability RepresentationLanguage ModelingLanguage ModelsSentiment AnalysisText RepresentationWord EmbeddingsBias In Language ModelsBias In Word EmbeddingsBiasesLanguage Bias Toward DisabilityMental Illness RepresentationNeural Language ModelsNeural Text RepresentationTextual Sentiment Analysis
Building equitable and inclusive NLP technologies demands consideration of whether and how social attitudes are represented in ML models. In particular, representations encoded in models often inadvertently perpetuate undesirable social biases from the data on which they are trained. In this paper, we present evidence of such undesirable biases towards mentions of disability in two different English language models: toxicity prediction and sentiment analysis. Next, we demonstrate that the neural embeddings that are the critical first step in most NLP pipelines similarly contain undesirable biases towards mentions of disability. We end by highlighting topical biases in the discourse about disability which may contribute to the observed model biases; for instance, gun violence, homelessness, and drug addiction are over-represented in texts discussing mental illness.
26
Cynthia Dwork, Moritz Hardt, Toniann Pitassi et al. · 2012 · 3.4K citations
A Synopsis of Linguistic Theory, 1930-1955
J. R. Firth · Medical Entomology and Zoology · 1957 · 1.7K citations