A Novel Algorithm of Internet Public Opinion Evaluation

Abstract

With the speedy interaction and spread of the network information, it is great significance to collect emerging massive internet information and discover hotspots of network public opinion. The paper presents a new model for evaluating internet public opinions based on improved BP neural network. First, a new evaluation indicator system containing 4 hierarchies and 26 third-grade indicators is designed based on the characteristics analysis of internet public opinions. Second, a new internet public opinion evaluation model is presented using genetic algorithm to speed up the convergence and simplify the model structure of the model, and the calculation steps of the model are designed. Finally the experimental results verify that the effectiveness and validity of the model can be guaranteed when used for evaluating internet public opinions practically.

DOI: http://dx.doi.org/10.11591/telkomnika.v11i7.2850