Generative Aspect-Based Sentiment Analysis with Contrastive Learning and Expressive Structure

TL;DR

This paper introduces GEN-SCL-NAT, a novel approach combining contrastive learning and a structured generation format to improve generative aspect-based sentiment analysis, especially for implicit sentiments, achieving state-of-the-art results.

Contribution

The paper proposes GEN-SCL-NAT, integrating supervised contrastive learning and a new structured generation format to enhance quadruple extraction in ACOS tasks, particularly for implicit expressions.

Findings

Achieves 1.48% average F1 improvement over baselines.

Significant gains on implicit aspect and opinion detection.

Top performance across three ACOS datasets.

Abstract

Generative models have demonstrated impressive results on Aspect-based Sentiment Analysis (ABSA) tasks, particularly for the emerging task of extracting Aspect-Category-Opinion-Sentiment (ACOS) quadruples. However, these models struggle with implicit sentiment expressions, which are commonly observed in opinionated content such as online reviews. In this work, we introduce GEN-SCL-NAT, which consists of two techniques for improved structured generation for ACOS quadruple extraction. First, we propose GEN-SCL, a supervised contrastive learning objective that aids quadruple prediction by encouraging the model to produce input representations that are discriminable across key input attributes, such as sentiment polarity and the existence of implicit opinions and aspects. Second, we introduce GEN-NAT, a new structured generation format that better adapts autoregressive encoder-decoder models to extract quadruples in a generative fashion. Experimental results show that GEN-SCL-NAT achieves top performance across three ACOS datasets, averaging 1.48% F1 improvement, with a maximum 1.73% increase on the LAPTOP-L1 dataset. Additionally, we see significant gains on implicit aspect and opinion splits that have been shown as challenging for existing ACOS approaches.