A Dynamic Human Contacts Prediction Method in Mobile Social Networks

Human contacts prediction is a challenging task in mobile social networks. In this paper, we extract some new features to predict human contacts and propose a new human contacts prediction method which is suitable for dynamic networks. The method first trains a classifier for each time period and assigns different weights to the classifiers of each time period. Then, an ensemble result of all the classifiers is used to predict human contacts. The experimental results show that our proposed method is efficient.