Learning Time-Varying Turn-Taking Behavior in Group Conversations

2025-10-21

AI Summary

Existing group dialogue turn-taking models lack cross-group generalizability and rely heavily on static, universal assumptions, failing to capture individual heterogeneity and dynamic behavioral evolution. To address this, we propose a flexible probabilistic framework that jointly encodes individual personality traits and historical utterance sequences. Crucially, it introduces, for the first time, a time-decaying “speaking propensity” mechanism—explicitly modeling how elapsed time since the last utterance modulates current speaking probability. The method integrates individual trait embeddings, sequential behavioral modeling, and an interpretable probabilistic formulation. We validate it on synthetic data and diverse real-world multi-party conversations. Experiments demonstrate substantial improvements in cross-group turn prediction accuracy. Moreover, our analysis reveals fundamental limitations of conventional static models in capturing temporal dependencies and individual-level variability in interactive dynamics. This work establishes a novel paradigm for modeling complex group interactions.

Technology Category

• Cognitive Modeling & Cognitive Systems: Social Cognition And Interaction
• Reasoning under Uncertainty: Relational Probabilistic Models
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Web Mining and Content Analysis: Models for Web evolution

Abstract

We propose a flexible probabilistic model for predicting turn-taking patterns in group conversations based solely on individual characteristics and past speaking behavior. Many models of conversation dynamics cannot yield insights that generalize beyond a single group. Moreover, past works often aim to characterize speaking behavior through a universal formulation that may not be suitable for all groups. We thus develop a generalization of prior conversation models that predicts speaking turns among individuals in any group based on their individual characteristics, that is, personality traits, and prior speaking behavior. Importantly, our approach provides the novel ability to learn how speaking inclination varies based on when individuals last spoke. We apply our model to synthetic and real-world conversation data to verify the proposed approach and characterize real group interactions. Our results demonstrate that previous behavioral models may not always be realistic, motivating our data-driven yet theoretically grounded approach.

Problem

• Predict turn-taking patterns in group conversations using individual characteristics
• Learn how speaking inclination varies based on when individuals last spoke
• Overcome limitations of universal models that cannot generalize across different groups

Innovation

• Probabilistic model predicts turn-taking in groups
• Learns speaking inclination from last speaking time
• Uses individual traits and past behavior data