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

Robustness of machine learning models on ever-changing real-world data iscritical, especially for applications affecting human well-being such ascontent moderation. New kinds of abusive language continually emerge in onlinediscussions in response to current events (e.g., COVID-19), and the deployedabuse detection systems should be updated regularly to remain accurate. In thispaper, we show that general abusive language classifiers tend to be fairlyreliable in detecting out-of-domain explicitly abusive utterances but fail todetect new types of more subtle, implicit abuse. Next, we propose aninterpretability technique, based on the Testing Concept Activation Vector(TCAV) method from computer vision, to quantify the sensitivity of a trainedmodel to the human-defined concepts of explicit and implicit abusive language,and use that to explain the generalizability of the model on new data, in thiscase, COVID-related anti-Asian hate speech. Extending this technique, weintroduce a novel metric, Degree of Explicitness, for a single instance andshow that the new metric is beneficial in suggesting out-of-domain unlabeledexamples to effectively enrich the training data with informative, implicitlyabusive texts.

Improving Generalizability in Implicitly Abusive Language Detection with Concept Activation Vectors