In this paper, we address the importance of sentiment analysis in extracting insights from textual data online. Traditional methods and deep learning models fall short in capturing complex relationships between entities, prompting us to introduce Relational Graph Convolutional Networks (RGCNs). This approach enhances interpretability and flexibility by capturing dependencies between data points as nodes in a graph. By utilizing pre-trained language models like BERT and RoBERTa with RGCN architecture on Amazon and Digikala product reviews, we showcase the efficiency of RGCNs in capturing relational information for sentiment analysis. This innovative method offers promising results in understanding user sentiments effectively.
https://arxiv.org/abs/2404.13079