Efficiency Improvement Of Mining The Region Of Interest On Moving Object

With the rapid development of the sensor network,radio communication technology,as well as the large-scale popularization of space orientation,people's ability to continuously monitor the spatial position of various kinds of moving objects has been enhanced,these massive trajectory data of high-speed growth revealed some important information about moving objects.It has become more and more important to research on region of interest discovery and location recommendation in trajectory data mining.It was found that when the classical DBSCAN in the procedure of the existing historical region of interest discovery was directly used for the clustering of the staypoints,the boundary points which belong to some classes of clusters were difficult to be accurately determined after studying systematically the theory and method of region of interest discovery and location recommendation in trajectory data mining.On the other hand,considering the hunting for the whole data set of the staypoints when iteration,the time cost of computation was very high.In addition,because of the widespread use of collaborative filtering based on object,the current potential region of interest recommendation was hard to achieve for poor time efficiency leading by complicated process.Aiming at the problems in the historical region of interest mining,an improved DBSCAN algorithm was designed and applied to the discovery of interest region on the basis of the classical DBSCAN,which gained good effects.The improved DBSCAN algorithm,based on the DBSCAN,introduced the minimum reachable distance in OPTICS,and utilized the minimum reachable distance of each data point to determine the cluster,and used the partition grid to improve the efficiency.Potential region of interest recommendation was analogical to item-based collaborative filtering recommendation algorithm,from the perspective of region,the target object was recommended to those areas that were similar to previously visited regions.The process was simpler than collaborative filtering based on object,which could improve the efficiency of recommendation.The experimental results showed that the improved DBSCAN has better cohesion and higher efficiency.The collaborative filtering recommendation based on region was less accurate,but the efficiency was much higher than that of object-based collaborative filtering.