UNF-BMI at SemEval-2026 Task 3: Research Domain Criteria-Guided Large Language Models for Dimensional Aspect-Based Sentiment Analysis

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Natural Language Processing · Depth: Expert, quick

Summary

The UNF-BMI system, developed for SemEval-2026 Task 3, Track A, Subtask 1 (Dimensional Aspect Sentiment Regression, DimASR), predicts continuous Valence-Arousal (VA) scores for specific aspects within text. This approach uniquely integrates psychologically grounded affective signals derived from the Research Domain Criteria (RDoC) framework. Two complementary methods are explored: an in-context learning framework utilizing Mistral-7B-Instruct with semantically retrieved few-shot examples, enhanced by lexicon-derived RDoC valence and arousal cues; and a supervised multi-task learning model based on RoBERTa, where VA regression is the primary objective and RDoC-based signal prediction serves as an auxiliary regularization task. Experiments on English laptop and restaurant review datasets demonstrate that RDoC-inspired affective priors effectively reduce Root Mean Square Error (RMSE) compared to baselines, particularly in text with sparse explicit sentiment cues.

Key takeaway

For NLP Engineers developing advanced sentiment analysis systems, you should consider integrating psychologically grounded frameworks like RDoC. This approach, demonstrated by the UNF-BMI system, significantly reduces RMSE in dimensional aspect-based sentiment regression, particularly when explicit sentiment cues are sparse. Implementing either in-context learning with LLMs augmented by RDoC cues or a multi-task RoBERTa model with RDoC-based auxiliary tasks can yield more accurate and robust Valence-Arousal predictions for your applications.

Key insights

Integrating RDoC-inspired affective signals into LLMs enhances dimensional aspect-based sentiment analysis, particularly for low-signal text.

Principles

• RDoC framework offers robust affective priors.
• Auxiliary tasks regularize shared representations.
• Psychological grounding aids low-signal text.

Method

In-context learning with Mistral-7B-Instruct uses RDoC cues. Multi-task RoBERTa performs VA regression with RDoC signal prediction as an auxiliary regularization task.

In practice

• Apply RDoC cues for VA score prediction.
• Use multi-task learning for robust models.
• Improve sentiment analysis in subtle texts.

Topics

• Dimensional Aspect Sentiment Analysis
• Valence-Arousal
• Research Domain Criteria
• Large Language Models
• Multi-task Learning
• SemEval-2026

Best for: Research Scientist, AI Scientist, NLP Engineer, Machine Learning Engineer

Related on AIssential

• Beyond Sentiment Classification: A Generative Framework for Emotion Intensity Evaluation in Text — Generative LLMs fine-tuned on continuous emotion intensity scores outperform classification and generalize to unseen affective dimensions.
• Learning Word Vectors for Sentiment Analysis: A Python Reproduction — Word vectors can effectively capture both semantic similarity and sentiment polarity by combining unsupervised and supervised learning objectives.
• Aspect-Based Sentiment Analysis in the Accommodation Domain Using the BERTimbau Model — BERTimbau model fine-tuning strategies for Portuguese ABSA show trade-offs between recall and precision.
• Consistency Analysis of Sentiment Predictions using Syntactic & Semantic Context Assessment Summarization (SSAS) — SSAS improves LLM sentiment prediction consistency by pre-processing noisy data into structured, high-signal prompts.
• From Prediction to Justification: Aligning Sentiment Reasoning with Human Rationale via Reinforcement Learning — Integrating explicit reasoning into sentiment analysis models enhances interpretability and performance by mimicking human cognitive processes.
• On GATE, Text and Social Media Analysis, and Detecting Misinformation Online — A multi-layer analytical pipeline effectively extracts and categorizes digital identity aspects of agrifood products from online discourse.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.