Bias Against 93 Stigmatized Groups in Masked Language Models and Downstream Sentiment Classification Tasks.
Katelyn MeiSonia FereidooniAylin CaliskanPublished in: FAccT (2023)
Keyphrases
- language model
- language modeling
- speech recognition
- n gram
- document retrieval
- probabilistic model
- document level
- language modelling
- sentiment analysis
- retrieval model
- context sensitive
- information retrieval
- test collection
- query expansion
- language model for information retrieval
- smoothing methods
- vector space model
- query terms
- statistical language models
- language models for information retrieval
- relevance model
- sentiment classification
- ad hoc information retrieval
- pseudo relevance feedback
- translation model
- document ranking
- term dependencies
- document length
- machine learning