Bayesian Spectral Emotion Transition Discovery from Multi-Annotator Disagreement

2026-06-01

AI Summary

This study addresses the limitations of conventional majority voting in emotion annotation, which overlooks inter-annotator disagreement and fails to capture dynamic emotional transitions in dialogues. The authors propose BSETD, a two-stage framework that first estimates a transition matrix with confidence intervals via hierarchical Dirichlet-Multinomial posterior inference while controlling the false discovery rate (FDR), and then decomposes emotional inertia (low-frequency) and contagion (high-frequency) components using symmetric graph Laplacian spectral decomposition. Innovatively modeling annotation uncertainty as an outer product of soft labels, the method integrates Bayesian inference with spectral graph theory to uncover, for the first time, transition patterns aligned with Plutchik’s adjacent emotions and Russell’s valence reversal. Experiments on EmotionLines reveal significant positive transitions between disgust and anger, and negative ones between joy and anger; cross-corpus validation across five sources shows high within-English correlations (0.91–0.98), moderate cross-lingual (Chinese–English) correlations (0.79–0.85), and a 0.979 correlation between human hard labels and LLM-derived soft labels.

Abstract

Emotions evolve through the dynamics of conversation, and understanding their transition structure is foundational to applications ranging from mental-health screening to dialogue systems. However, existing studies typically compress multi-rater judgments into a single hard label by majority voting, discarding the uncertainty signal needed to understand turn-to-turn transitions. In this article, we propose Bayesian Spectral Emotion Transition Discovery (BSETD), a two-stage framework that discovers emotion-transition structure from multi-rater soft labels. In the first stage, a hierarchical Dirichlet-Multinomial posterior is constructed through the outer product of soft labels, equipping each cell of the K x K transition matrix with a credible interval and Benjamini-Hochberg (BH) false discovery rate (FDR)-controlled significance. In the second stage, the symmetrized graph Laplacian is spectrally decomposed to separate a low-frequency (inertia) component from a high-frequency (contagion) component. On EmotionLines, BSETD simultaneously recovers the signatures of two distinct affective spaces: the Plutchik-adjacent transitions disgust to anger (log2 lift +0.94) and anger to disgust (+0.86) are over-represented, while the Russell-valence-reversed transitions joy to anger (-0.90) and anger to joy (-0.89) are under-represented. A five-source cross-corpus validation yields pairwise Pearson correlations in 0.91-0.98 within English, 0.79-0.85 against Chinese M3ED, and 0.979 between the human hard labels and the LLM virtual soft labels on the same utterance set, demonstrating that a pipeline preserving annotator uncertainty bridges the computational study of emotion dynamics with established psychological theory.

Problem

• emotion transition
• multi-annotator disagreement
• uncertainty
• conversation dynamics
• soft labels

Innovation

• Bayesian inference
• emotion dynamics
• multi-annotator disagreement
• spectral decomposition
• soft labels