Research On Sentiment Analysis Model Based On Hybrid Neural Network Introducing Attention Mechanism

Natural language processing uses the extraction of sentiment information in comment texts to conduct sentiment analysis research,but traditional sentiment analysis methods have the problems of insufficient feature extraction,sparse data,high dimensionality,and poor generalization ability.In order to solve the above problems,this paper adopts the sentiment analysis method based on deep learning from two different perspectives of model fusion and feature fusion,and introduces attention mechanism to conduct the study of sentiment analysis.First of all,in order to overcome the problems of gradient disappearance and convolutional neural network ignoring context information in the recurrent neural network,this paper uses a combination of a bidirectional long and short-term neural network and a convolutional neural network to overcome the disadvantages of using a single neural network;It is shared with the parameters in the convolutional neural network without highlighting the status quo of the keyword’s effect on sentiment analysis.An attention mechanism is introduced to reduce the interference of noise and fully extract the global and local features of the text.Secondly,the existing sentiment analysis methods only focus on the global text information,and ignore the different degrees of user preference and product performance impact on sentiment analysis.This article is based on the combination of two-way gated recursive unit and convolutional neural network.In the above,the user attention mechanism and product attention mechanism are introduced,the user features and product features are fully extracted,and the user features and product features are merged to obtain the final text features for sentiment analysis.Finally,design experiments to verify the two sentiment analysis models proposed in this paper,and analyze the experimental results in detail.