A unified dual-view knowledge-guided sentiment interaction networks for aspect-based sentiment analysis.

Area of Science:

• Natural Language Processing
• Artificial Intelligence
• Machine Learning

Background:

• Aspect-based sentiment analysis (ABSA) identifies sentiment towards specific entities and attributes.
• Existing methods struggle to effectively integrate external knowledge for augmenting sequence semantics and capturing sentiment-syntax relationships.

Purpose of the Study:

• To propose the Dual-view Knowledge Guided Sentiment Interaction Network (Dual-KGIN) to address limitations in ABSA.
• To enhance both sentiment semantics and syntactic representations by integrating external knowledge and hierarchical interactions.

Main Methods:

• Developed an external knowledge-guided syntactic Graph Convolutional Network (GCN) module to refine dependencies and enhance syntactic features.
• Augmented sequence semantics using external knowledge with an attention mechanism.
• Introduced a multi-level feature interaction module for improved sentiment representation.

Main Results:

• Dual-KGIN demonstrated superior performance in aspect-specific sentiment identification on benchmark ABSA datasets.
• Ablation studies confirmed the effectiveness of individual components within the Dual-KGIN framework.

Conclusions:

• Dual-KGIN effectively integrates external knowledge and models feature interactions for state-of-the-art ABSA.
• The proposed network offers a unified framework for enhancing sentiment semantics and syntactic representations.