A Multi-Dimensional Text Sentiment Analysis Method Based on Joint Network

Abstract

Text sentiment analysis in social media has problems such as irregular structure, short length and sparse features. In this paper, a text sentiment analysis method combining emotional symbols is proposed. Based on the BiGRU and capsule network joint network model, this method fully considers the influence of emotional emoji in the text to be analyzed on the sentiment analysis tendency. Secondly, the BiGRU network was used to extract the long-term dependent features of the text context, and the capsule network was used to deal with the problem of losing feature information in the CNN pooling layer to better extract the local features of the text. Finally, the Softmax classifier was used to output the sentiment tendency. Experimental results show that the proposed model is superior to the current mainstream models in accuracy, recall rate and F1 value.