New TIDES dataset enhances AI's understanding of multi-party conversations

Researchers have introduced TIDES, a new longitudinal dataset designed to improve the modeling of multi-party social dynamics in group conversations. This dataset comprises over 75,000 utterances in English and Korean from 12 university project teams over a full semester, offering a naturalistic view of team evolution. Experiments show that fine-tuning models on TIDES enhances next-speaker prediction and performs comparably to proprietary models on benchmarks like the AMI Meeting Corpus, though human evaluations indicate that improved prediction doesn't always lead to more natural-sounding dialogue. AI

IMPACT Improves AI's ability to model complex group conversations and team dynamics.

RANK_REASON The cluster contains a research paper detailing a new dataset and experimental results. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• alphaXiv
• AMI Meeting Corpus
• arXiv
• CatalyzeX
• DagsHub
• English
• Gotit.pub
• Hugging Face
• Korean
• ScienceCast
• TIDES

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →