Application of a multi-channel attention mechanism in text classification of new media

Sentiment analysis of new media text has become a research hotspot in recent years. In order to more effectively analyse the emotional polarity of new media text, this paper proposes a text classification algorithm based on a multi-channel attention mechanism. First, channels based on bidirectional gating recurrent unit (BiGRU) neural network are used to extract semantic features, while channels based on fully connected neural network are used to extract emotional features. In order to extract the key information better, the attention mechanism is introduced into the two channels respectively, and the bidirectional encoder representation technology based on converter is used to provide the word vectors. Then, the real emotional semantics are embedded into the model through the dynamic adjustment of the word vector by the context. Finally, the semantic features and emotional features of the double channels are fused to obtain the final semantic expression. In the experiment part, NLPCC2014 dataset and microblog data captured by crawler are used for comparative experiment. The experimental results show that the proposed multi-channel attention mechanism method can enhance the ability of emotion semantic capture, and improve the performance of emotion classification.