LogSigma at SemEval-2026 Task 3: Uncertainty-Weighted Multitask Learning for Dimensional Aspect-Based Sentiment Analysis

TL;DR

LogSigma is a novel multitask learning system that uses uncertainty weighting to improve continuous sentiment prediction across languages, achieving top results in DimABSA at SemEval-2026.

Contribution

Introduces a method that automatically balances regression tasks in DimABSA using learned uncertainty, tailored for multilingual and multi-domain settings.

Findings

Achieved 1st place on five datasets across tracks.

Uncertainty weights vary significantly across languages.

Language-specific task difficulty profiles influence optimal weighting.

Abstract

This paper describes LogSigma, our system for SemEval-2026 Task 3: Dimensional Aspect-Based Sentiment Analysis (DimABSA). Unlike traditional Aspect-Based Sentiment Analysis (ABSA), which predicts discrete sentiment labels, DimABSA requires predicting continuous Valence and Arousal (VA) scores on a 1-9 scale. A central challenge is that Valence and Arousal differ in prediction difficulty across languages and domains. We address this using learned homoscedastic uncertainty, where the model learns task-specific log-variance parameters to automatically balance each regression objective during training. Combined with language-specific encoders and multi-seed ensembling, LogSigma achieves 1st place on five datasets across both tracks. The learned variance weights vary substantially across languages due to differing Valence-Arousal difficulty profiles-from 0.66x for German to 2.18x for English-demonstrating that optimal task balancing is language-dependent and cannot be determined a priori.