From Affect Prediction to Affect Forecasting: Evidence for Distinct Information Sources in Longitudinal Text

2026-06-15

AI Summary

This study distinguishes between the predictive mechanisms for current emotional states and future emotional changes in longitudinal textual data, revealing that they rely on distinct information sources: current emotion is primarily driven by textual semantics, whereas emotional change depends more heavily on historical numerical trajectories. To address this, the authors propose E-TSAP, a temporal semantic affect prediction model for current emotion estimation, and introduce ACF-Hybrid, a hybrid model incorporating dimension-specific numerical features for forecasting subsequent emotional shifts. Experimental results demonstrate that E-TSAP achieves correlation coefficients of r = 0.670 for valence and r = 0.449 for arousal, while ACF-Hybrid attains r ≈ 0.658 in predicting emotional change—significantly outperforming text-only baselines. This work provides the first empirical evidence supporting the adoption of distinct modeling paradigms for these two related yet fundamentally different tasks.

Abstract

Modeling dimensional affect in longitudinal text requires distinguishing current affect estimation from future affective change forecasting. Existing approaches often treat each text as an independent observation and apply similar assumptions to both tasks, without testing whether they rely on different information sources. This paper investigates that distinction using longitudinal self-reported ecological essays and feeling-word entries. We propose the Trait--State Affective Prediction (TSAP) framework and its temporal extension E-TSAP for per-text valence and arousal prediction, evaluated on a held-out prediction test set of 1,737 entries from 91 users. We further propose the Affective Change Forecaster Hybrid (ACF-Hybrid) for next-step affective change forecasting, evaluated on a held-out forecasting test set of 46 users. For prediction, E-TSAP achieves composite Pearson correlations of 0.670 for valence and 0.449 for arousal. For forecasting, textual representations perform worse than compact numeric trajectory baselines: the text-inclusive model achieves only r=0.316 for valence and r=0.284 for arousal, whereas a simple prior-state baseline reaches r=0.615 and r=0.670, respectively. ACF-Hybrid, using dimension-specific numeric trajectory features, achieves r=0.659 for valence and $r=0.658$ for arousal. These results show that textual semantics support current affect prediction, whereas future affective change is better captured through prior numeric trajectory dynamics.

Problem

• affect prediction
• affect forecasting
• longitudinal text
• dimensional affect
• information sources

Innovation

• affect forecasting
• trait-state modeling
• longitudinal text
• affective change
• temporal dynamics