Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AILS-NTUA at SemEval-2026 Task 3: Efficient Dimensional Aspect-Based Sentiment Analysis

Stavros Gazetas, Giorgos Filandrianos, Maria Lymperaiou, Paraskevi Tzouveli, Athanasios Voulodimos, Giorgos Stamou

TL;DR

AILS-NTUA system for Track-A of SemEval-2026 Task 3 on Dimensional Aspect-Based Sentiment Analysis (DimABSA) is presented, which encompasses three complementary problems: Dimensional Aspect Sentiment Regression, Dimensional Aspect Sentiment Triplet Extraction, and Dimensional Aspect Sentiment Quadruplet Prediction within a multilingual and multi-domain framework.

Abstract

In this paper, we present AILS-NTUA system for Track-A of SemEval-2026 Task 3 on Dimensional Aspect-Based Sentiment Analysis (DimABSA), which encompasses three complementary problems: Dimensional Aspect Sentiment Regression (DimASR), Dimensional Aspect Sentiment Triplet Extraction (DimASTE), and Dimensional Aspect Sentiment Quadruplet Prediction (DimASQP) within a multilingual and multi-domain framework. Our methodology combines fine-tuning of language-appropriate encoder backbones for continuous aspect-level sentiment prediction with language-specific instruction tuning of large language models using LoRA for structured triplet and quadruplet extraction. This unified yet task-adaptive design emphasizes parameter-efficient specialization across languages and domains, enabling reduced training and inference requirements while maintaining strong effectiveness. Empirical results demonstrate that the proposed models achieve competitive performance and consistently surpass the provided baselines across most evaluation settings.

AILS-NTUA at SemEval-2026 Task 3: Efficient Dimensional Aspect-Based Sentiment Analysis

TL;DR

AILS-NTUA system for Track-A of SemEval-2026 Task 3 on Dimensional Aspect-Based Sentiment Analysis (DimABSA) is presented, which encompasses three complementary problems: Dimensional Aspect Sentiment Regression, Dimensional Aspect Sentiment Triplet Extraction, and Dimensional Aspect Sentiment Quadruplet Prediction within a multilingual and multi-domain framework.

Abstract

In this paper, we present AILS-NTUA system for Track-A of SemEval-2026 Task 3 on Dimensional Aspect-Based Sentiment Analysis (DimABSA), which encompasses three complementary problems: Dimensional Aspect Sentiment Regression (DimASR), Dimensional Aspect Sentiment Triplet Extraction (DimASTE), and Dimensional Aspect Sentiment Quadruplet Prediction (DimASQP) within a multilingual and multi-domain framework. Our methodology combines fine-tuning of language-appropriate encoder backbones for continuous aspect-level sentiment prediction with language-specific instruction tuning of large language models using LoRA for structured triplet and quadruplet extraction. This unified yet task-adaptive design emphasizes parameter-efficient specialization across languages and domains, enabling reduced training and inference requirements while maintaining strong effectiveness. Empirical results demonstrate that the proposed models achieve competitive performance and consistently surpass the provided baselines across most evaluation settings.
Paper Structure (43 sections, 6 equations, 4 figures, 12 tables)

This paper contains 43 sections, 6 equations, 4 figures, 12 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Task and Dataset Description
  4. Task Definition
  5. Dataset Description
  6. Evaluation Metrics
  7. DimASR
  8. DimASTE and DimASQP
  9. System Overview
  10. Aspect-Based Regression for DimASR
  11. Language backbones.
  12. Instruction-Tuned LLM Generation for DimASTE and DimASQP
  13. Unified prompting with and without categories.
  14. Structured outputs and VA handling.
  15. Backbones and parameter-efficient fine-tuning.
  16. ...and 28 more sections

Figures (4)

  • Figure 1: DimASR dataset statistics across splits: number of reviews (left) and aspects per review (right).
  • Figure 2: DimASTE and DimASQP dataset statistics across splits: number of reviews (left) and quadruplets per review (right).
  • Figure 3: DimASTE and DimASQP NULL analysis on Train set: NULL rate (left) and composition of NULL cases (right).
  • Figure 4: DimASTE and DimASQP distribution shift measured by PSI between Train and Dev/Test across features (review length, quadruplets/review, and category).