Design And Implementation Of News Public Opinion Analysis System Based On Neural Network

Living in the Internet age,a growing number of people participate in online life to express their ideas,opinions and attitudes anytime and anywhere.At present,the number of Internet users in China has reached more than 1 billion.The ability of the network to reflect the real society has been greatly enhanced,and online public opinion may have a far-reaching impact on individuals and even enterprises.Any negative reports or remarks on the Internet may have an impact on the products and reputation of an enterprise,or even bring about huge losses of economic interests.In view of the above background,this thesis studies the literature related to entity-based emotion analysis,compares the advantages and disadvantages of different implementations,then designs and implements an entity-based emotion classification model,SATAE-TC-LSTM,which introduces syntactic analysis into the network and combines with attention mechanism.Finally,the accuracy of this method is improved in the entity sentiment classification by compared experiments,and it has a good effect in the entity sentiment classification of Chinese.So it can be applied to sentiment analysis based on different entities in news.Based on the above sentiment analysis algorithm based on specific entities,this thesis designs and implements a news public opinion analysis system.Based on Express,Vue and Flask frameworks,the system implements web services from news’s collection,preprocessing,analysis and report’s generation.Finally,the system is tested.In functional testing,the system meets the functional needs of enterprise users and analysts.In non-functional testing,the system meets the needs of users for system security,and has good reliability and interaction.