English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Groundbreaking inventions and highly significant performance improvements indeep learning based Natural Language Processing are witnessed through thedevelopment of transformer based large Pre-trained Language Models (PLMs). Thewide availability of unlabeled data within human generated data deluge alongwith self-supervised learning strategy helps to accelerate the success of largePLMs in language generation, language understanding, etc. But at the same time,latent historical bias/unfairness in human minds towards a particular gender,race, etc., encoded unintentionally/intentionally into the corpora harms andquestions the utility and efficacy of large PLMs in many real-worldapplications, particularly for the protected groups. In this paper, we presentan extensive investigation towards understanding the existence of "AffectiveBias" in large PLMs to unveil any biased association of emotions such as anger,fear, joy, etc., towards a particular gender, race or religion with respect tothe downstream task of textual emotion detection. We conduct our exploration ofaffective bias from the very initial stage of corpus level affective biasanalysis by searching for imbalanced distribution of affective words within adomain, in large scale corpora that are used to pre-train and fine-tune PLMs.Later, to quantify affective bias in model predictions, we perform an extensiveset of class-based and intensity-based evaluations using various biasevaluation corpora. Our results show the existence of statistically significantaffective bias in the PLM based emotion detection systems, indicating biasedassociation of certain emotions towards a particular gender, race, andreligion.

Blacks is to Anger as Whites is to Joy? Understanding Latent Affective Bias in Large Pre-trained Neural Language Models