KAIST releases TIDES: a semester-long bilingual dataset of real team collaboration
josephseering · x · 2026-09-11
TIDES: A Longitudinal Bilingual Dataset for Multi-Party Social Dynamics
A KAIST team (COLM 2026) released TIDES, a dataset for modeling social dynamics in multi-party conversation. The researchers followed 12 real student teams over a full semester to capture how collaboration and social dynamics develop over time.
Scale:
- 12 teams (7 Korean / 5 English), 88 meetings, 104 transcripts
- 75,971 utterances in total
Addressing three limitations of existing datasets:
- Scripted/lab/synthetic dialogues → TIDES uses in-the-wild recordings of actual course projects
- Single short sessions → TIDES covers a full semester longitudinally
- Fixed assigned roles (e.g., professor–student) → TIDES annotates emergent, behavior-grounded roles: dominance, sociability, and task orientation
The dataset offers three annotation layers linking utterances, emergent roles, and team development stages across each team's timeline. Available on Hugging Face, with GitHub and arXiv versions.
More from Research
- Tencent Hunyuan releases T1: 122B MoE model RL-trained for long-horizon terminal tasks — Tencent-Hunyuan · 2026-09-11
- BOTANIC-1 model family launches to design tomorrow's food from DNA up, fully open on Hugging Face — CatAstro_Piyush · 2026-09-11
- MSRA, UTS and Tsinghua unveil UniSteer: injecting human corrections into RL for flow-matching VLAs — jiqizhixin · 2026-09-11
- World Model RL Debiasing Cuts Cost of Scaling Autonomous Research Agents — illinois · 2026-09-11
- Links: Economist AI-labor article and adaptive capacity paper overview — soumitrashukla9 · 2026-09-11
- FrankenSim: A Pure-Rust Unified Physics & Design Kernel Built to Be Driven by AI Agents — doodlestein · 2026-09-11