Boundary-Driven Table-Filling with Cross-Granularity Contrastive Learning for Aspect Sentiment Triplet Extraction

2025-02-04

AI Summary

ASTE requires joint extraction of aspect terms, opinion terms, and their associated sentiment polarities. Existing end-to-end two-dimensional table-filling methods over-rely on token-level interactions while neglecting sentence-level global semantics, limiting their ability to model multi-token aspects/opinions and complex syntactic structures. To address this, we propose a boundary-driven two-dimensional table-filling framework that pioneers a modeling paradigm integrating sentence-level global representations with token-level local interactions. We introduce cross-granularity contrastive learning to align semantic representations across sentence and token levels, and design a multi-scale convolutional module to enhance multi-granular feature capture. Extensive experiments on multiple benchmark datasets achieve state-of-the-art F1 scores, demonstrating significant improvements in robustness and accuracy for extracting multi-token entities and capturing long-range dependencies.

Technology Category

• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Knowledge Representation and Reasoning: Argumentation
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Sentiment analysis and opinion mining

Abstract

The Aspect Sentiment Triplet Extraction (ASTE) task aims to extract aspect terms, opinion terms, and their corresponding sentiment polarity from a given sentence. It remains one of the most prominent subtasks in fine-grained sentiment analysis. Most existing approaches frame triplet extraction as a 2D table-filling process in an end-to-end manner, focusing primarily on word-level interactions while often overlooking sentence-level representations. This limitation hampers the model's ability to capture global contextual information, particularly when dealing with multi-word aspect and opinion terms in complex sentences. To address these issues, we propose boundary-driven table-filling with cross-granularity contrastive learning (BTF-CCL) to enhance the semantic consistency between sentence-level representations and word-level representations. By constructing positive and negative sample pairs, the model is forced to learn the associations at both the sentence level and the word level. Additionally, a multi-scale, multi-granularity convolutional method is proposed to capture rich semantic information better. Our approach can capture sentence-level contextual information more effectively while maintaining sensitivity to local details. Experimental results show that the proposed method achieves state-of-the-art performance on public benchmarks according to the F1 score.

Problem

• Extract aspect, opinion terms, and sentiment polarity
• Enhance semantic consistency across granularity levels
• Improve global context capture in complex sentences

Innovation

• Boundary-driven table-filling technique
• Cross-granularity contrastive learning method
• Multi-scale convolutional semantic capture