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

Neural abstractive summarization models are prone to generate summaries whichare factually inconsistent with their source documents. Previous work hasintroduced the task of recognizing such factual inconsistency as a downstreamapplication of natural language inference (NLI). However, state-of-the-art NLImodels perform poorly in this context due to their inability to generalize tothe target task. In this work, we show that NLI models can be effective forthis task when the training data is augmented with high-quality task-orientedexamples. We introduce Falsesum, a data generation pipeline leveraging acontrollable text generation model to perturb human-annotated summaries,introducing varying types of factual inconsistencies. Unlike previouslyintroduced document-level NLI datasets, our generated dataset contains examplesthat are diverse and inconsistent yet plausible. We show that models trained ona Falsesum-augmented NLI dataset improve the state-of-the-art performanceacross four benchmarks for detecting factual inconsistency in summarization.  The code to obtain the dataset is available online athttps://github.com/joshbambrick/Falsesum

Falsesum: Generating Document-level NLI Examples for Recognizing Factual Inconsistency in Summarization