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

Contrast consistency, the ability of a model to make consistently correctpredictions in the presence of perturbations, is an essential aspect in NLP.While studied in tasks such as sentiment analysis and reading comprehension, itremains unexplored in open-domain question answering (OpenQA) due to thedifficulty of collecting perturbed questions that satisfy factualityrequirements. In this work, we collect minimally edited questions aschallenging contrast sets to evaluate OpenQA models. Our collection approachcombines both human annotation and large language model generation. We findthat the widely used dense passage retriever (DPR) performs poorly on ourcontrast sets, despite fitting the training set well and performingcompetitively on standard test sets. To address this issue, we introduce asimple and effective query-side contrastive loss with the aid of dataaugmentation to improve DPR training. Our experiments on the contrast setsdemonstrate that DPR's contrast consistency is improved without sacrificing itsaccuracy on the standard test sets.

Exploring Contrast Consistency of Open-Domain Question Answering Systems on Minimally Edited Questions