一种基于VARdnn-Tree的反向最近邻查询方法

Method of reverse nearest neighbor queries using VARdnn-Tree

空间数据库中反向最近邻查询在低维查询时一般利用基于R-Tree的改进树作为索引结构,由于树型索引结构本身的限制,R-Tree等索引结构的查询在高维中都会出现维数灾难。针对这个问题,提出了一种基于VARdnn-Tree的索引结构,采用量化压缩的方法存储数据,能够有效地支持高维查询。

In spatial databases, the improved R-Tree is usually used as indexing structure in the research of low-dimension reverse nearest neighbor queries. Queries by indexing structures such as R-Tree will cause dimension disaster in high-dimension because of the limitation of tree indexing structure. To solve the problem, this paper proposed an indexing structure based on VARdnn-Tree. It adoped the method of the quantification compression to store data, thus could effectively support queries in high-dimension.