In order to solve the problem that the existing link prediction models based on local information between nodes considered the dependent relationships between common neighbor nodes insufficiently and failed to fully make use of the network topology information, meanwhile improve the accuracy of links prediction, this paper put forward the link prediction method based on hidden naive Bayes model. The algorithm fully considered the interdependence between common neighbor nodes and difference between interdependence. Then the similarity of nodes were computed through hidden naive Bayes classification model and the dependence between nodes were measured by utilizing the conditional mutual information. Through the above methods, the link prediction accuracy was finally improved. In the simulation, DBLP and Email data sets were used as the experimental data and the method of AUC and Precision were used to evaluate the forecasting models. Results show that the predictive effect of proposed algorithm is better than that of the mainstream method which effectively verified the accuracy of the method.
