基于加权Voronoi图的top-k局部同位模式挖掘OA

Local co-location pattern(LCP)mining is an important branch of the spatial co-location pattern mining,which aims to discover co-location patterns that prevalently occur in local regions.The LCPs can reveal the association relation-ships between spatial features in local regions rather than globally,playing a positive guiding role in various location-based application fields.Existing LCP mining methods cannot effectively identify local regions formed by human activities(human factors),and it is difficult to set an appropriate prevalence threshold to select prevalent patterns in different regions.To address these problems,a novel top-k LCP mining method based on weighted Voronoi diagram(Top-k LCPM-WVD)is proposed.This method first identifies the distribution regions of LCPs formed by human factors through the weighted Voronoi diagram,and then uses a top-k mining framework to efficiently mine the k most prevalent patterns in each region.At the same time,a series of optimization strategies is designed based on this framework to further improve the mining efficiency.In addition,in order to solve the efficiency problem when facing large scale datasets,a parallel mining scheme is proposed to speed up the mining process,achieving a speedup ratio of 1.65 under 4 threads.Extensive experimental results on both real and synthetic datasets demonstrate that the proposed Top-k LCPM-WVDmethod more efficiently discovers interpretable LCPs compared with existing state-of-the-art algorithms,with efficiency improvements up to several dozen times.