FairTalk: Facilitating Balanced Participation in Video Conferencing by Implicit Visualization of Predicted Turn-Grabbing Intention

2025-05-26

AI Summary

Unequal speaking opportunities in video conferencing critically undermine collaborative fairness. To address this, we propose an implicit intervention method: (1) a turn-taking intention prediction model built via positive-unlabeled (PU) learning—enabling real-time intent recognition without requiring explicit negative labels for interruptions; and (2) anthropomorphic micro-gesture visual feedback (e.g., subtle nodding, gaze aversion) that subtly rebalances speaking rights without disrupting conversational flow. Our approach integrates turn-taking detection, behavior-informed visual design heuristics, and empirical user studies. Experimental results demonstrate a statistically significant improvement in speaking-turn distribution balance (p < 0.01). While subjective perception did not reach statistical significance, qualitative feedback strongly supports the intervention’s effectiveness and design feasibility. This work establishes a novel paradigm for non-intrusive, interpretable enhancement of fairness in remote meetings.

Technology Category

• Humans and AI: Interaction Techniques and Devices
• Computer Vision: Video Understanding & Activity Analysis
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Economics, Online Markets and Human Computation: Fairness, privacy, and diversity in economic environments

Abstract

Creating fair opportunities for all participants to contribute is a notable challenge in video conferencing. This paper introduces FairTalk, a system that facilitates the subconscious redistribution of speaking opportunities. FairTalk predicts participants' turn-grabbing intentions using a machine learning model trained on web-collected videoconference data with positive-unlabeled learning, where turn-taking detection provides automatic positive labels. To subtly balance speaking turns, the system visualizes predicted intentions by mimicking natural human behaviors associated with the desire to speak. A user study suggests that FairTalk may help improve speaking balance, though subjective feedback indicates no significant perceived impact. We also discuss design implications derived from participant interviews.

Problem

• Balancing speaking opportunities in video conferencing
• Predicting turn-grabbing intentions using machine learning
• Visualizing intentions to improve participation fairness

Innovation

• Machine learning predicts turn-grabbing intentions
• Visualizes intentions via natural human behaviors
• Uses positive-unlabeled learning for training