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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Large language models (LLMs) are rapidly changing various domains. However,their capabilities in handling conversational breakdowns still require anin-depth exploration. This paper addresses the challenge of detecting andmitigating dialogue breakdowns within LLM-driven conversational systems. Whilepowerful models from OpenAI and Anthropic excel in many dialogue tasks, theycan still produce incoherent or contradictory responses, commonly referred toas breakdowns, which undermine user trust. To tackle this, we propose anapproach that combines specialized fine-tuning with advanced promptingstrategies, including few-shot learning, chain-of-thought reasoning, andanalogical prompting. In particular, we fine-tune a small 8B model anddemonstrate its robust classification and calibration capabilities in Englishand Japanese dialogue. We also validate its generalization on the BETOLDdataset, achieving a 7\% accuracy improvement over its base model. Furthermore,we introduce a real-time deployment architecture that selectively escalatessuspicious responses to more resource-intensive frontier models only whenbreakdowns are detected, significantly cutting operational expenses and energyconsumption. Experimental results show our method surpasses priorstate-of-the-art specialized classifiers while also narrowing performance gapsbetween smaller open-source models and large proprietary ones. Our approachoffers a scalable solution for robust conversational AI in high-impact domainsby combining efficiency, interpretability, and reliability.

Towards Robust Dialogue Breakdown Detection: Addressing Disruptors in Large Language Models with Self-Guided Reasoning