Bayesian Spectral Emotion Transition Discovery from Multi-Annotator Disagreement
2026-06-01 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors address how emotions change during conversations by using the full range of multiple people's uncertain labels instead of simplifying them to one fixed label. They introduce a two-step method called BSETD that builds a transition matrix with uncertainty estimates and then breaks down emotional shifts into common (inertia) and influential (contagion) parts. Testing on conversation data shows their method matches known psychological patterns and is reliable across languages and label types. This approach helps better understand how emotions flow in dialogue by keeping multiple opinions instead of losing information through majority voting.
emotion transitionmulti-rater soft labelsBayesian modelingDirichlet-Multinomialfalse discovery ratespectral decompositiongraph Laplacianaffective spacesemotion dynamicscross-corpus validation
Authors
Keito Inoshita, Takato Ueno
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.