Aspect Information Enhanced Contrastive Learning for Aspect-based Sentiment Analysis
Abstract: Aspect-based sentiment analysis (ABSA) is a fine-grained sentiment classification task. Despite significant improvements in this field, progress is hindered by challenges such as the scarcity of context for specific aspects, interference from irrelevant words in sentences, and a lack of research focus on leveraging correlations between samples. To address these issues, we introduce a novel method named Aspect Information Enhanced Contrastive Learning Learning (AIECL) for ABSA. Firstly, we employ cutting-edge prompting techniques with Large Language Models (LLMs) to generate nuanced aspect-specific descriptions, thereby enhancing contexts related to the aspect. Subsequently, we design a novel fusion module aimed at seamlessly amalgamating aspectual insights with the original sentence structure. Finally, we develop three pioneering contrastive learning strategies aimed at exploring and learning complex correlations between samples, which is crucial for fine-grained sentiment analysis.
Experiments on six benchmark datasets demonstrate that our AIECL method substantially outperforms state-of-the-art techniques and provides valuable insights for applying LLMs to downstream tasks.
Paper Type: Long
Research Area: Sentiment Analysis, Stylistic Analysis, and Argument Mining
Research Area Keywords: Sentiment Analysis; Large Lanugage Models; Contrastive Learning;
Contribution Types: NLP engineering experiment
Languages Studied: English
Submission Number: 2527
Loading