Construction Of Network Public Opinion Warning Model Based On BP And Elman Neural Network

In today's Internet age,the Internet has become a particularly important part of people's daily lives,both in life and in work.It brings a lot of convenience,but also brings a lot of network sensation troubles to this aera,even affects the harmony and stability of the society.Therefore,there are more and more researches on how to warn the network public opinion crisis in the current network age,and build a harmonious and healthy It is very necessary to see the network environment.The research mainly divides the network public opinion crisis warning into two research parts.The first is to analyze the sentiment tendency of the collected sample data,and the second is to establish the public opinion early warning crisis model.This paper is about the alysis of cyber lyric sentiment based on Doc2 vec and Bi-LSTM neural networks,which combines the expression attention mechanism.Taking into account the interaction between emoji and text,and finally the overall sentimental.Secondly,based on BP and Elman neural network algorithm,the warning model of public opinion crisis is established,and the two neural network models are trained,tested,verified and compared,and finally a feasible algorithm model is obtained.The main work of this paper is:1.Introduced the warning model index system established by individual research,and listed the previous network public opinion crisis warning related algorithm and its algorithm application.2.In this paper,the Doc2 vec model and the Bi-LSTM neural network model are combined with the emoji attention mechanism to analyze the sentiment orientation of the data.The Doc2 vec model is used to generate the word vector,then the Bi-LSTM neural network makes good use of the context feature.The information,which retains the order information of the text,can automatically select the feature as the classification,and combine the emoji weights,and finally obtain the emotional category probabilities to analyze the sentiment orientation through the function Softmax.3.An warning index system is established for the public opinion warning model,which mainly analyzes and divides the indicators from four levels of indicators: event force,netizen force,media influence,and government guidance.Create BP and Elman neural network models,and extract sample data from the microblog topic “Chongqing bus crashing into the river”.Two neural network models were trained and tested separately.And compare the prediction error of the two model algorithms to verify the feasibility and effectiveness of the neural network,so as to obtain a better warning effect of the network public opinion crisis.