Structural Attention Neural Networks for improved sentiment analysis
This work addresses sentiment analysis for natural language processing applications, representing an incremental improvement over existing recursive models.
The authors tackled sentiment classification by introducing a tree-structured attention neural network that incorporates structural information and attention mechanisms, achieving state-of-the-art performance on the Stanford Sentiment Treebank dataset.
We introduce a tree-structured attention neural network for sentences and small phrases and apply it to the problem of sentiment classification. Our model expands the current recursive models by incorporating structural information around a node of a syntactic tree using both bottom-up and top-down information propagation. Also, the model utilizes structural attention to identify the most salient representations during the construction of the syntactic tree. To our knowledge, the proposed models achieve state of the art performance on the Stanford Sentiment Treebank dataset.