Label-Consistent Data Generation for Aspect-Based Sentiment Analysis Using LLM Agents

TL;DR

This paper introduces an agentic data augmentation approach for Aspect-Based Sentiment Analysis that iteratively generates and verifies synthetic data, improving label accuracy and model performance over traditional prompting methods.

Contribution

The paper presents a novel agentic augmentation method that outperforms prompting-based baselines in generating high-quality, label-consistent synthetic data for ABSA tasks.

Findings

Agentic augmentation yields higher label preservation than prompting.

Combining augmented data with real data improves model performance.

Method benefits are more significant with less pretrained models.

Abstract

We propose an agentic data augmentation method for Aspect-Based Sentiment Analysis (ABSA) that uses iterative generation and verification to produce high quality synthetic training examples. To isolate the effect of agentic structure, we also develop a closely matched prompting-based baseline using the same model and instructions. Both methods are evaluated across three ABSA subtasks (Aspect Term Extraction (ATE), Aspect Sentiment Classification (ATSC), and Aspect Sentiment Pair Extraction (ASPE)), four SemEval datasets, and two encoder-decoder models: T5-Base and Tk-Instruct. Our results show that the agentic augmentation outperforms raw prompting in label preservation of the augmented data, especially when the tasks require aspect term generation. In addition, when combined with real data, agentic augmentation provides higher gains, consistently outperforming prompting-based generation. These benefits are most pronounced for T5-Base, while the more heavily pretrained Tk-Instruct exhibits smaller improvements. As a result, augmented data helps T5-Base achieve comparable performance with its counterpart.