Adaptive Data Augmentation for Aspect Sentiment Quad Prediction
Wenyuan Zhang, Xinghua Zhang, Shiyao Cui, Kun Huang, Xuebin Wang, Tingwen Liu
TL;DR
This work tackles data imbalance in aspect sentiment quad prediction by introducing Adaptive Data Augmentation (ADA) and a Knowledge-aware Generator. It formalizes quad-pattern and aspect-category imbalances, employs DAG-based representations and a condition-function-guided concatenation augmentation to balance data, and enriches augmented samples with category priors and a semantic, constraint-based target decoding. Empirical results on four public ASQP benchmarks show state-of-the-art performance, with detailed ablations confirming the contributions of adaptive augmentation and knowledge-aware decoding. The approach offers practical improvements for ABSA systems and suggests future integration with large language models to further handle open-world and reasoning aspects.
Abstract
Aspect sentiment quad prediction (ASQP) aims to predict the quad sentiment elements for a given sentence, which is a critical task in the field of aspect-based sentiment analysis. However, the data imbalance issue has not received sufficient attention in ASQP task. In this paper, we divide the issue into two-folds, quad-pattern imbalance and aspect-category imbalance, and propose an Adaptive Data Augmentation (ADA) framework to tackle the imbalance issue. Specifically, a data augmentation process with a condition function adaptively enhances the tail quad patterns and aspect categories, alleviating the data imbalance in ASQP. Following previous studies, we also further explore the generative framework for extracting complete quads by introducing the category prior knowledge and syntax-guided decoding target. Experimental results demonstrate that data augmentation for imbalance in ASQP task can improve the performance, and the proposed ADA method is superior to naive data oversampling.