TIDES: A Longitudinal Bilingual Dataset for Modeling Multi-Party Social Dynamics

2026-08-03

AI Summary

This study addresses the challenge that current large language models struggle to capture the dynamics of long-term, multi-party group conversations in real-world settings, primarily due to a lack of high-resolution, longitudinal, naturally occurring multilingual dialogue data. To bridge this gap, we introduce the TIDES dataset, which tracks face-to-face meetings of 12 university teams over an entire semester, comprising 75,971 English–Korean bilingual utterances annotated for interaction types, emergent roles, and developmental stages. Leveraging this resource, we propose a fine-grained social structure annotation framework and fine-tune large language models for next-speaker prediction. Experimental results show that our TIDES-finetuned model achieves 64.53% accuracy—13.8 percentage points higher than a dyadic baseline—and approaches state-of-the-art performance on the AMI corpus while using 42% less training data, demonstrating its effectiveness in modeling authentic team dynamics.

Abstract

Group conversations are fundamental to human collaboration, yet standard large language models (LLMs) still struggle with the complexities of multi-party interaction. This challenge persists in part because existing group conversation datasets are often limited to short-term lab settings with contrived tasks, failing to capture the long-term social dynamics of real-world teams. To bridge this gap, we introduce TIDES, a high-resolution longitudinal dataset tracking 12 university project teams over a full semester. Comprising 75,971 utterances in both English and Korean from in-person meetings, TIDES provides a naturalistic record of teams working on self-managed projects. Our socio-structural annotations-covering interaction types, emergent roles, and development stages-allow for modeling of team evolution over months. Experiments show that fine-tuning on TIDES improves next-speaker prediction by 13.8 percentage points over a bigram baseline (64.53%) and yields performance comparable to strong proprietary zero-shot models. The model also comes within 2.1 percentage points of the published state of the art on the AMI Meeting Corpus while using approximately 42% less training data. However, human evaluations suggest that better next-speaker prediction does not necessarily yield more natural or coherent utterances, as fine-tuned models were generally less preferred than vanilla models. This potential mismatch motivates further study of how structural modeling can support natural multi-party generation.

Problem

• multi-party conversation
• social dynamics
• longitudinal dataset
• team interaction
• next-speaker prediction

Innovation

• longitudinal dataset
• multi-party conversation
• next-speaker prediction
• socio-structural annotation
• bilingual dialogue