Research On Multi-Source Nearest Neighbor Query Method In Road Network Environment

With the development of mobile internet technology,the smooth deployment and implementation of a large number of applications depend on queries of spatial data objects to varying degrees.Spatial nearest neighbor query is one of the main types of queries related to spatial objects,which is used to find the nearest target object to users.However,the current nearest neighbor queries have some problems,such as single query source,scattered and disordered query results,too large result set and can not meet user preferences.To solve the above problems,this paper mainly studies multi-source Nearest Neighbor queries,which are used to return target objects satisfying spatial and attribute constraints to avoid the shortcomings of single-source nearest neighbor queries.The multi-source Nearest Neighbor queries can be applied not only to location analysis,tourism planning,logistics distribution and other daily life services,but also to locationrelated spatial entity object query and analysis.Firstly,aiming at the single query source of existing k NN queries,the nearest neighbor query method,which consists of multiple query points,is studied by using spatial distance constraint as the main condition of query.Based on the idea of node extension of two single-source basic algorithms(Dijkstra and IER),multi-source nearest neighbor query algorithms based on local priority strategy and global priority strategy are proposed,R-tree index is constructed and compact upper and lower bounds are designed,which make the access scope of nodes reduced and the performance of the algorithm improved.Secondly,aiming at the problem that the existing nearest neighbor query results are scattered and disorderly,and the target objects? location are clustered together(such as shopping malls,factory buildings,warehouses,etc.),the research on multi-source nearest neighbor queries with common constraints of spatial distance and attribute keywords is studied,and the method of calculating the influence of target group and the algorithm of keyword query are proposed.The algorithm aggregates the target objects according to the set rules,merges the target objects within the distance of the user-specified road network into a group,and forms a road network containing multiple target groups.After transformation,the search of big-scale with small targets in the original road network is transformed into the search of small-scale with big targets,which reduces the number of nodes and meets the needs of aggregated queries.In the transformed road network,the matching analysis between the overall attributes of the target group(the union of all the attributes of the target points)and the query keywords input by users is carried out,and the target group which satisfies the full coverage of attributes and has great influence is formed into a group set and returned to users.Finally,aiming at the problem that the existing k NN query result sets are too large to satisfy user preferences,the optimization methods that are applicable to both aggregated and non-aggregated query results are studied.The basic idea of multiobjective optimization method "Skyline technology" is referred,the nearest neighbor result set is analyzed by dominant relation,and the dominant target points in dominant relation are screened out,which are divided into designated size point groups to form an orderly set of point groups.In order to select the Top-k results in the result set,an optimization method based on Layer Skyline is proposed.According to the characteristics of Skyline low-level points dominating the high-level points,according to the number of low-level points contained in the target point group,a better group of points in the domination relationship is selected.In order to deal with the point group of equivalent relation formed in the optimization result of Layer Skyline,an optimization method based on Group Skyline is proposed.In the point group of layer equivalence,the number of dominant points of each target object in the point group is calculated,and the advantages and disadvantages among the point groups of layer equivalence are further judged,and the ordered result set of user preference is formed.