Food Safety Network Public Opinion Analysis Based On Deep Learning

In recent years,frequent food safety incidents have caused consumers to become extremely uneasy.At the same time,with the rapid development and rise of the Internet,more and more people are participating in the sensational discussion through the Internet.It is of great practical significance to understand the public opinion through the Internet,pay attention to public opinion trends,and determine the people's attitude toward public opinion,in order to understand and guide the Internet public opinion,ensure the healthy development of the food industry,and promote social harmony and stability.The text representation in the Internet public opinion analysis is the basis of all subsequent analysis.At present,the text representation mainly uses the word-based model.However,such models cannot directly represent the documents,which makes it impossible to represent unregistered words.To solve this problem,this paper proposes a document representation model WADM(Wasserstein Adversarial Document Model)based on the GAN(Generative Adversarial Network)network in deep learning.The model uses two neural networks to learn from each other and uses self-encoded encoders.As its discriminating network,the hidden layer directly obtains the distribution of the document,and has stronger document representation ability than the word-based model.Based on the text representation,this paper studies the text sentiment analysis algorithm.The existing text sentiment analysis algorithm is mainly based on rules and statistical machine learning algorithms.It requires manual design of rules and features,which is cumbersome.In response to this problem,this paper uses the LSTM(Long Short Term Memory)and CNN(Convolutional Neural Network)deep learning techniques to propose a network sentiment orientation algorithm C-LSTM(Convolutional Long).Short Term Memory)uses LSTM to process text data,and features feature extraction and sentiment classification by CNN.It solves the problem of complicated manual design rules and features,and improves classification accuracy.Finally,this paper realizes the whole process of obtaining and analyzing food security from the food safety network,collecting public opinion data from various channels,and using the WADM and C-LSTM algorithms proposed in this paper.The experimental results show that the algorithm can effectively represent the lyrics text and better classify the emotional attitudes of netizens in the topic.