Mining Prevalent Co-location Patterns Based On Global Topological Relations

With the rapid popularity of mobile devices and the Internet,everyone is producing spatial data all the time,so people's lives are filled with huge amounts of spatial data.Big data are far from us in the eyes of ordinary people,but actually considerable data are very close to us.How to obtain people's unknown knowledge from massive spatial data and provide services for people's lives is a very important problem at present.There are many sources and types of spatial data,including map data,image data,terrain data,attribute data and metadata.Obviously,the characteristics of spatial data,such as positioning,qualitative,temporal and spatial relations,make it difficult for traditional data analysis technology to process immense spatial data directly.Therefore,the exploration of efficient spatial data mining technology is an important goal of data mining researchers.Spatial co-location pattern mining is an extremely important research direction in the spatial data mining area.A spatial co-location pattern is a subset of spatial features whose instances are frequently located together in the geographic space.The proximity between instances is determined by a unified distance threshold given by the user in traditional spatial co-location pattern mining.However,the proximity defined by a unified distance threshold has many disadvantages.First,the user doesn't know which distance threshold is best suited in most cases,even for experts.Secondly,different densities of instance distribution are not considered in the same dataset and it cannot obtain complete proximity relationships when using a unified distance threshold to measure the proximity.Thirdly,the global topological relations of instances are ignored in mining and the mining results are not accurate.In order to solve the above-mentioned problems,this paper first introduces some knowledge of the Delaunay triangulation and spatial co-location pattern mining.Secondly,the method of calculating the distance constraint of instance is proposed and the spatial proximity is redefined according to the distance constraint.The new proximity solves the problem that it is very difficult for the user to give the distance threshold.On the basis of the new definition of spatial proximity relationship,a novel data structure for storing the adjacent relations between instances,called the proximity relationship tree,is proposed.According to the proximity relationship tree,a framework for mining prevalent co-location patterns based on the global topological relations,is proposed.Then,a prevalent co-location pattern mining algorithm based on the global topological relations is designed.Finally,the mining effectiveness and efficiency of the algorithm are evaluated on real data sets and simulated data sets.