Analysis Of Similarity About Trajectory Of User Behavior Based On Check_in Data

Spatio-temporal trajectory is recorded sequences of the location and time about moving objects.As a kind of important temporal object data types and sources of information,scope of application of the spatio-temporal trajectory covers many aspects such as user behavior,intelligent transportation and precision marketing and so on.With the rapid development of satellite positioning technology,wireless communication,intelligent terminals and mobile Internet,people can more easily obtain trajectory data.Through analysis on all kinds of spatio-temporal trajectory data,we can extract the similarity characteristics of spatio-temporal trajectory data,without priori knowledge,having similar behavior of spatio-temporal objects into together,and having a different behavior of the spatiotemporal objects to separate,whose key is according to the characteristics of spatio-temporal trajectory data,designing and defining the method of similarity measure between different spatio-temporal trajectory.The main research contents of this paper include:Because of location-based social networks,users sign only after reaching a certain position,no continuously tracking trajectory of user behavior,and user sign behavior with some random and repeatability,resulting large difference in different position registration number,a few users accomplishing the most attendance and some positions being rarely sign and unequal dimension of unequal making user check-in data showing the sparse.This paper,by check-in time divided into several time periods,use the optics algorithm hierarchical clustering based on density for check-in points of interest of users,getting user regions of interest in different time division under different scales of the spatial division,which is more reasonable than those of the grid or a single space division scale,reflecting the distribution of users’ spatio-temporal data better.At the same time,the similarity of each region of interest is compared with the similar bounding box,which is more in line with the characteristics of the check-in data,which greatly reduces the computational complexity,and the computational efficiency is improved.In addition,spatio-temporal trajectory are divided from the time dimension,which can adjust the weight of each time period according to the specific application,so as to reflect the important degree of the check-in data in each time period.Finally,this paper uses the user check-in data in the Gowalla of the large social network,and the results of the experiment are evaluated by the index of precison and recall rate.