基于时序特征的移动模式挖掘

Mining mobility patterns based on temporal and sequential features

定位设备(如GPS)的广泛使用产生了大量时空轨迹,合理地利用这些轨迹数据可以帮助挖掘用户移动模式.本文基于轨迹的时序特征提出一种新的模型来挖掘用户隐含移动模式.考虑到轨迹的特点:(1)位置顺序对于理解用户移动模式很重要;(2)用户的移动模式在不同时间段存在变化,本文提出的方法首先对位置序列进行建模,然后将时间信息加入到模型中.为了验证模型的有效性,本文在Gowalla签到数据集上进行了详细的实验,实验结果表明提出的模型优于传统的LDA模型及T-Bi LDA模型.

The wide-spread use of positioning devices（e.g., GPS） has given rise to a mass of spatio-temporal trajectories, which enable us to mine user-mobility patterns. In this paper, we proposed a model based on sequential and temporal trajectory features to mine people＇s latent movement patterns. Considering the following trajectory characteristics—（1） location sequences play a pivotal role in understanding user-mobility patterns, and（2） user-mobility patterns change over time—we first modeled the location sequences and then incorporated the temporal information into the model. To verify the effectiveness of our model, we performed thorough empirical studies on a check-in dataset of the Gowalla social network. The experimental results confirmed that the proposed method performed better than Latent Dirichlet Allocation（LDA） and T-Bi LDA.