Modeling and analyzing a public opinion influence method with K-adaboost

Abstract

In order to solve the low efficiency of public opinion influence analysis of social media, a new public opinion influence algorithm K-adaboost has been proposed in this paper according to adaboost and K-means algorithms. We first group the training samples and calculate the clustering center of all types of users in the group using the K-means algorithm, and then train the weak classifier of public opinion data and confirm the influence of public opinion on all types of users using the adaboost algorithm, so as to get the total influence of public opinions. Finally, we compare and analyze the performance of K-adaboost, K-means and adaboost algorithms through simulation experiments. The results show that K-adaboost has good adaptability in convergence time and accuracy.