ICT-NLP at SemEval-2026 Task 3: Less Is More -- Multilingual Encoder with Joint Training and Adaptive Ensemble for Dimensional Aspect Sentiment Regression

TL;DR

This paper presents a resource-efficient multilingual encoder system for Dimensional Aspect Sentiment Regression, utilizing joint training, a bounded regression method, and adaptive ensemble strategies, achieving top rankings in SemEval-2026.

Contribution

It introduces a lightweight multilingual system with joint training and adaptive ensemble techniques, avoiding reliance on large language models or external data.

Findings

Achieved top rankings on multiple datasets in SemEval-2026.

Demonstrated strong and consistent performance across languages.

Improved training stability with a bounded regression approach.

Abstract

This paper describes our system to SemEval-2026 Task 3 Track A Subtask 1 on Dimensional Aspect Sentiment Regression (DimASR). We propose a lightweight and resource-efficient system built entirely on multilingual pre-trained encoders, without relying on LLMs or external corpora. We adopt joint multilingual and multi-domain training to facilitate cross-lingual transfer and alleviate data sparsity, introduce a bounded regression transformation that improves training stability while constraining predictions within the valid range, and employ an adaptive ensemble strategy via subset search to reduce prediction variance. Experimental results demonstrate that our system achieves strong and consistent performance, ranking 1st on zho-res, 2nd on zho-lap, and 3rd on jpn-hot, with all remaining datasets placed within the top half of participating teams.