Research On Evaluation And Early Warning Algorithm Of Food Safety Network Public Opinion

With the rapid development of the Internet,self-media has also developed rapidly,and network public opinion has received more and more attention from the society.Because food safety concerns the national economy and people’s livelihood,food safety network public opinion has always been social hot spot.Especially in recent years,as the terribly increased events of the various food safety network public opinion,it has attracted the great attention of the state.The food safety network public opinion evaluation and early warning is an important part of public opinion analysis.This paper studies the evaluation and early warning method of food safety network public opinion,and constructs a food safety network public opinion index system and public opinion evaluation and early warning model,which is significance for the steady development of the food industry and social harmony.Reasonable and measurable index system is the foundation and primary condition of constructing evaluation and early warning model of food safety network public opinion,the current evaluation and early warning index system of the food safety internet public opinion is not comprehensive to consider the theme attribute,the dissemination and diffusion index,and does not consider the self-characteristics and evolution law of public opinion in depth.Aiming at the above problems,willing add theme attribute and propagation and diffusion index as first level index to public opinion index system,and constructing an index system of five dimensions,including theme attribute,source of public opinion,content elements of public opinion,audience of public opinion,propagation and diffusion.Among,the theme attribute index can better grasp the characteristics and background of public opinion events,and the propagation and diffusion index can excavate evolvement and process of public opinion.Based on reasonable construction of the index system,this paper studies the evaluation and early warning algorithm of food safety network public opinion.Existing network public opinion evaluation and early warning algorithms have shortcomings in characteristics and the long-term dependence of data sequence,and may occur over-fitting and gradient disappearance during training.Aiming at the above problems,based on LSTM(Long Short Term Memory)and regularization method,constructing Re-LSTM(Regularization LSTM)early warning model of food safety network public opinion.The current evaluation model of network public opinion is too traditional and single.This paper constructs BiLSTM(Binary LSTM)evaluation model of food safety network public opinion based on the LSTM,and conducts experiments and analysis.The experimental results show that the model constructed in this time can not only improve the accuracy of evaluation and early warning,but also better avoid the problems of gradient disappearance and over-fitting.Finally,developing a food safety network public opinion evaluation and early warning system,which realizes the whole process from data collection to data analysis,collects micro-blog and news public opinion data,and conducts food safety network public opinion evaluation and early warning based on Re-LSTM and BiLSTM algorithms,and acquires good effect.