Recurrent Convolutional Neural Network with Attention for Twitter and Yelp Sentiment Classification

Abstract

This paper proposes a combination structure of Recurrent Neural Network (RNN) and Convolutional Neural Network (CNN) for sentiment classification. Our model has two distinct characteristics: (1) Owing to the characteristics of the combination structure, our model can extract n_gram features with bidirectional time series. (2) Our model utilizes the attention mechanism to effectively capture important emotional content. Experiments conducted on two public datasets twitter and yelp. We evaluate the two aspects of this paper and shows that our model has a significant performance.