MASTE: A Multi-Agent Pipeline for Zero-Shot Aspect Sentiment Triplet Extraction

2026-07-08

AI Summary

This work addresses the challenge of accurately extracting aspect term, opinion term, and sentiment polarity triplets from review sentences in zero-shot settings, where large language models often struggle. To overcome this limitation, the authors propose a multi-agent pipeline architecture that decomposes the aspect sentiment triplet extraction (ASTE) task into four sequential stages, each handled by a specialized agent. The agents perform explicit, conditional reasoning based on the outputs of preceding stages, enabling effective zero-shot collaborative inference without any training. This approach pioneers the use of sequential conditional generation with multiple agents for ASTE, eliminating reliance on labeled examples or handcrafted prompts. Evaluated on four benchmark datasets, the method significantly outperforms existing zero-shot and chain-of-thought baselines, substantially narrowing the performance gap with fully supervised models while demonstrating strong generalization across both models and datasets.

Abstract

Aspect Sentiment Triplet Extraction (ASTE) requires jointly identifying (aspect, opinion, sentiment) triples from a given review sentence. While large language models (LLMs) achieve strong zero-shot performance on many NLP benchmarks, their effectiveness on ASTE remains limited, as single-pass generation forces the model to determine span boundaries, opinion grouping, and sentiment polarity in a single decoding step. Common remedies, such as few-shot in-context learning and chain-of-thought prompting, offer only marginal improvements and rely heavily on either in-domain demonstrations sampled from labeled training data or carefully engineered reasoning prompts, neither of which is broadly available in zero-shot deployment. Inspired by the classical agent paradigm, we propose MASTE, a multi-agent pipeline for zero-shot Aspect Sentiment Triplet Extraction. MASTE decomposes ASTE into four sequential stages, where specialized agents handle different compositional subtasks with explicit conditioning on prior outputs. This design enables entirely training-free zero-shot ASTE and generalizes across different backbones and datasets. Extensive experiments on four ASTE benchmarks show that MASTE substantially outperforms zero-shot and chain-of-thought LLM baselines under the same backbone, narrowing the gap to fully supervised methods without using any labeled triplets. Code is available at https://github.com/Hankerlove/MASTE.

Problem

• Aspect Sentiment Triplet Extraction
• zero-shot learning
• large language models
• sentiment analysis
• natural language processing

Innovation

• Multi-Agent System
• Zero-Shot Learning
• Aspect Sentiment Triplet Extraction
• Prompt Engineering
• Large Language Models