Multi-layer GCNs for Travel Text Aspect-level Sentiment Classification
Min Wang, Ming Yang · 2023
Aspect category based travel text sentiment analysis is to extract the polarity and perspectives of sentimental travel experiences and feelings expressed by tourists on social media or tourism applications, in order to understand their satisfaction with scenic spots, tourism products and services. This paper proposes a tourism text category sentiment classification model based on hierarchical graph convolution to address the inherent relationship between the two subtasks of category detection and sentiment analysis that cannot be clearly modeled in traditional models. In this model, multiple aspect categories in a sentence are first detected, and then text sentiment features are obtained through two different graph convolutional networks. Then, sentiment information for each category is obtained through pooling output. Finally, the final category sentiment is obtained by combining the category detection results. The experimental results show that the proposed method outperforms baseline models such as Add-BERT, H-BERT, etc. in terms of accuracy, recall, and F1 value evaluation indicators.