Research On Target Mining Algorithm For Mass Location Information

With the development of positional technology and the popularity of smart mobile terminals,location-based services have been developed and applied rapidly,and hence a great deal of location information has been generated.The in-depth analyzing and mining of massive location information has become a better trend to implement location-based services.Therefore,it has very good practical significance and practical value.There are many difficulties in analysis and mining of massive location information,such as difficulties in storing massive location information,high computational capability requirements,difficulties of data mining analysis and results of such complicated issues.Aiming at these,we first studied the location information pre-processing technology,which is based on the position information of the characteristics of temporal and spatial rules improved density clustering algorithm,realizies trajectory clustering and noise elimination;and then study Z order curve of the space filling curves,which divides the plane of the earth into grids,and design the algorithm for calculating VC encoding and neighborhood grid;we then study the hot region based on VC encoding on mass position information extraction,hot path mining,sequential pattern mining,accompanying pattern mining,mining analysis,and propose the focus regional clustering algorithm,based on VC codes FP-growth and PrefixSpan algorithm data generation method,accompanying pattern mining;Finally,the algorithms have been validated and evaluated by using GPS location data of taxi in San Francisco.The experimental results show that the proposed mass location information mining algorithm has better mining efficiency when using with distributed database and Spark data mining platform,the running time of the algorithm scales with the data size,which means our proposed algorithms are capable of mass location information mining task.