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

The Counter Narrative (CN) is a promising approach to combat online hatespeech (HS) without infringing on freedom of speech. In recent years, there hasbeen a growing interest in automatically generating CNs using natural languagegeneration techniques. However, current automatic CN generation methods mainlyrely on expert-authored datasets for training, which are time-consuming andlabor-intensive to acquire. Furthermore, these methods cannot directly obtainand extend counter-knowledge from external statistics, facts, or examples. Toaddress these limitations, we propose Retrieval-Augmented Unsupervised CounterNarrative Generation (RAUCG) to automatically expand external counter-knowledgeand map it into CNs in an unsupervised paradigm. Specifically, we firstintroduce an SSF retrieval method to retrieve counter-knowledge from themultiple perspectives of stance consistency, semantic overlap rate, and fitnessfor HS. Then we design an energy-based decoding mechanism by quantizingknowledge injection, countering and fluency constraints into differentiablefunctions, to enable the model to build mappings from counter-knowledge to CNswithout expert-authored CN data. Lastly, we comprehensively evaluate modelperformance in terms of language quality, toxicity, persuasiveness, relevance,and success rate of countering HS, etc. Experimental results show that RAUCGoutperforms strong baselines on all metrics and exhibits strongergeneralization capabilities, achieving significant improvements of +2.0% inrelevance and +4.5% in success rate of countering metrics. Moreover, RAUCGenabled GPT2 to outperform T0 in all metrics, despite the latter beingapproximately eight times larger than the former. Warning: This paper maycontain offensive or upsetting content!

RAUCG: Retrieval-Augmented Unsupervised Counter Narrative Generation for Hate Speech