基于模式匹配度的用户移动规则挖掘及位置预测方法研究

Method of mining user mobile rule based on pattern matching degree and location prediction

传统的移动用户位置预测方法由于模式支持度计算方式不合理,存在预测精度偏低的问题。为此,提出了一种基于模式匹配度的用户移动规则挖掘及位置预测方法,并将其用于移动通信系统中,以基台覆盖范围网格为单元的用户位置预测。具体包括三个步骤:通过图的遍历挖掘用户移动模式、基于用户移动模式生成用户移动规则和依据用户移动规则进行位置预测。实验分析使用10个批次轨迹数据进行用户移动规则挖掘,结果表明,该方法挖掘出的用户移动规则数少、支持度高、置信度高,具有高精度的优点。

The traditional mobile user location prediction method has a problem of low prediction accuracy due to unreasonable calculation of support.To overcome these challenges,this paper presented a method called mining user mobile rule based on pattern matching degree and location prediction.It can predict location with the base coverage mesh as the unit for mobile users in mobile communication systems.The detailed procedure contained three phases:a)mining user mobile patterns from graph traversal;b)generating user mobile rules from user mobile patterns;c)predicting based on user mobile rules.Finally,the conducted experiments on 10 batches of trajectory data demonstrate that this method has the characteristics of fewer rules,higher support,and higher confidence than traditional methods,and highly predict accuracy.