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

In recent years, the availability of large-scale annotated datasets, such asthe Stanford Natural Language Inference and the Multi-Genre Natural LanguageInference, coupled with the advent of pre-trained language models, hassignificantly contributed to the development of the natural language inferencedomain. However, these crowdsourced annotated datasets often contain biases ordataset artifacts, leading to overestimated model performance and poorgeneralization. In this work, we focus on investigating dataset artifacts anddeveloping strategies to address these issues. Through the utilization of anovel statistical testing procedure, we discover a significant associationbetween vocabulary distribution and text entailment classes, emphasizingvocabulary as a notable source of biases. To mitigate these issues, we proposeseveral automatic data augmentation strategies spanning character to wordlevels. By fine-tuning the ELECTRA pre-trained language model, we compare theperformance of boosted models with augmented data against their baselinecounterparts. The experiments demonstrate that the proposed approacheseffectively enhance model accuracy and reduce biases by up to 0.66% and 1.14%,respectively.

Dissecting vocabulary biases datasets through statistical testing and  automated data augmentation for artifact mitigation in Natural Language  Inference