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

Incorporating external knowledge into dialogue generation (KIDG) is crucialfor improving the correctness of response, where evidence fragments serve asknowledgeable snippets supporting the factual dialogue replies. However,introducing irrelevant content often adversely impacts reply quality and easilyleads to hallucinated responses. Prior work on evidence retrieval andintegration in dialogue systems falls short of fully leveraging existingevidence since the model fails to locate useful fragments accurately andoverlooks hidden evidence labels within the KIDG dataset. To fully Unleash thepotential of evidence, we propose a framework to effectively incorporateEvidence in knowledge-Intensive Dialogue Generation (u-EIDG). Specifically, weintroduce an automatic evidence generation framework that harnesses the powerof Large Language Models (LLMs) to mine reliable evidence veracity labels fromunlabeled data. By utilizing these evidence labels, we train a reliableevidence indicator to effectively identify relevant evidence from retrievedpassages. Furthermore, we propose an evidence-augmented generator with anevidence-focused attention mechanism, which allows the model to concentrate onevidenced segments. Experimental results on MultiDoc2Dial demonstrate theefficacy of evidential label augmentation and refined attention mechanisms inimproving model performance. Further analysis confirms that the proposed methodoutperforms other baselines (+3~+5 points) regarding coherence and factualconsistency.

Unleashing Potential of Evidence in Knowledge-Intensive Dialogue Generation