A time online evaluation method over campus network log

Abstract

Internet addiction is a serious social problem nowadays. People with internet addiction usually spend much time on line, which there is no effectively technical way to evaluate. We propose a time-online evaluating method based on net log. The method gathers adjacent timestamps of every account's network log into a cluster, which can be seen as a time period. The sum of all periods is the account's total time online. Besides, we apply parallel processing in the data analysis and cluster. The experiments on real datasets show that this method is effective for time online evaluation.