Mining and Prediction of the Core Group in the Communication Network Based on Walk Trap and ARIMA Algorithm

Fulian Yin, Xiaowei Liu, Ge Su, Sitong Li

Abstract: In the existing researches, mining and prediction of the core group in the communication network are difficult to be correlated. In addition, the focus group for monitoring events cannot be concentrated. In order to solve these problems, this paper researches the mining and prediction method of the core group based on Walk Trap and ARIMA algorithm. It proposes an optimized core group mining method combining the degree centrality and Walk Trap algorithm, then puts forward the flagged line graph to connect the core group mining results and the subsequent core group forecast. In this way, the coverage of focus groups can be effectively identified so as to facilitate the communication network monitoring. The experiment results show that the method reduces the focus group to 13.03% compared with the coarse-grained coverage method, and avoids the omission of 4,167 key users compared with the subjective discrimination method. It greatly improves the monitoring efficiency and accuracy.