🤖 AI Summary
This paper addresses word-sense ambiguity induced by contextual information in aspect-level sentiment analysis. We propose a knowledge-enhanced dynamic modeling framework that explicitly incorporates the synonymy structure of knowledge graphs into this task for the first time. Our method jointly leverages BERT-derived semantic representations and knowledge graph embeddings, introduces a knowledge-driven state vector generation mechanism, and models aspect–sentiment temporal dependencies via a position-aware memory bank integrated with a gated recurrent unit (DCGRU). Key innovations include: (i) a knowledge-guided dynamic attention mechanism that refines contextual representations using external lexical knowledge; and (ii) a position-enhanced memory modeling framework that synergistically optimizes semantic knowledge integration and sequential dependency modeling. Experimental results on three benchmark datasets—SemEval-2014 Task 4, Twitter, and Rest14—achieve new state-of-the-art F1 scores, with an average improvement of 2.3% over prior methods.
📝 Abstract
In this paper, we propose a novel method to enhance sentiment analysis by addressing the challenge of context-specific word meanings. It combines the advantages of a BERT model with a knowledge graph based synonym data. This synergy leverages a dynamic attention mechanism to develop a knowledge-driven state vector. For classifying sentiments linked to specific aspects, the approach constructs a memory bank integrating positional data. The data are then analyzed using a DCGRU to pinpoint sentiment characteristics related to specific aspect terms. Experiments on three widely used datasets demonstrate the superior performance of our method in sentiment classification.