Data-Efficient Adaptation and a Novel Evaluation Method for Aspect-based Sentiment Analysis

2025-11-04

AI Summary

Aspect-Based Sentiment Analysis (ABSA) research lags in low-resource domains such as education and healthcare, hindered by scarce annotated data and rigid exact-match evaluation protocols. Method: We propose FTS-OBP—a flexible evaluation framework integrating text similarity and optimal bipartite matching—and conduct the first systematic investigation of efficient adaptation of small decoder-only generative models (1.5–3.8B parameters) to low-resource ABSA. Leveraging in-context learning, weight fusion, lightweight fine-tuning, and multi-task training, our approach achieves state-of-the-art performance on merely 200–1,000 labeled samples using a single GPU. Contribution/Results: We release the first publicly available ABSA dataset for educational reviews. Experiments demonstrate substantial improvements in domain generalization and evaluation validity, establishing a new paradigm for practical ABSA deployment in low-resource settings.

Technology Category

• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models
• Machine Learning: Evaluation and Analysis
• Application Domains: Humanities & Computational Social Science

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Web Mining and Content Analysis: Large pretrained models with web data
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Aspect-based Sentiment Analysis (ABSA) is a fine-grained opinion mining approach that identifies and classifies opinions associated with specific entities (aspects) or their categories within a sentence. Despite its rapid growth and broad potential, ABSA research and resources remain concentrated in commercial domains, leaving analytical needs unmet in high-demand yet low-resource areas such as education and healthcare. Domain adaptation challenges and most existing methods'reliance on resource-intensive in-training knowledge injection further hinder progress in these areas. Moreover, traditional evaluation methods based on exact matches are overly rigid for ABSA tasks, penalising any boundary variations which may misrepresent the performance of generative models. This work addresses these gaps through three contributions: 1) We propose a novel evaluation method, Flexible Text Similarity Matching and Optimal Bipartite Pairing (FTS-OBP), which accommodates realistic extraction boundary variations while maintaining strong correlation with traditional metrics and offering fine-grained diagnostics. 2) We present the first ABSA study of small decoder-only generative language models (SLMs;<7B parameters), examining resource lower bounds via a case study in education review ABSA. We systematically explore data-free (in-context learning and weight merging) and data-light fine-tuning methods, and propose a multitask fine-tuning strategy that significantly enhances SLM performance, enabling 1.5-3.8 B models to surpass proprietary large models and approach benchmark results with only 200-1,000 examples on a single GPU. 3) We release the first public set of education review ABSA resources to support future research in low-resource domains.

Problem

• Addressing domain adaptation challenges in low-resource ABSA
• Developing flexible evaluation methods for boundary variations
• Enabling data-efficient training for small generative language models

Innovation

• Novel FTS-OBP evaluation method for flexible boundary matching
• Data-light multitask fine-tuning for small generative language models
• First ABSA study on sub-7B models with minimal training data