**2. Model**

The architecture of the proposed SDNN model is shown in Figure 1, which consists of four key components: the feature-enhanced word-embedding module, bidirectional GRU network module, convolutional neural network module, and the sentence classifier module. In addition, both the bidirectional GRU network module and the convolutional neural network module jointly constitute the deep neural network module. The goal of our model is to predict the sentiment polarity of the given sentence. In the rest of this section, we will elaborate our SDNN model in detail.

**Figure 1.** The architecture of the proposed sentiment-feature-enhanced deep neural network (SDNN) model.
