Reducing Bias in Sentiment Analysis Models Through Causal Mediation Analysis and Targeted Counterfactual Training

Reducing Bias in Sentiment Analysis Models Through Causal Mediation Analysis and Targeted Counterfactual Training

Large language models provide high-accuracy solutions in many natural language processing tasks. In particular, they are used as word embeddings in sentiment analysis models. However, these models pick up on and amplify biases and social stereotypes in the data. Causality theory has recently driven...

Full description

Bibliographic Details
Main Authors: Yifei Da, Matias Nicolas Bossa, Abel Diaz Berenguer, Hichem Sahli
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Access
Subjects:
BERT
bias
causal mediation analysis
counterfactual training
large language models
sentiment analysis
Online Access:https://ieeexplore.ieee.org/document/10388308/

Internet

https://ieeexplore.ieee.org/document/10388308/

Similar Items