A fast construction method for spatial index GBD tree

This paper proposes a fast initial construction method of the GBDtree. The GDB tree has proper characteristics for management of large amount of 2 or 3 dimensional data. However, the GBDtree needs long initial construction time by originally proposed onebyone insertion method. A fast insertion method has been proposed, but it needs large size of buffer capable to hold index information of all entries. The paper proposes another fast initial construction method. The method requires only limited size of work space (buffer). The experimental results show the initial construction time reduces into a third or a quarter of the onebyone insertion method. The memory efficiency and retrieval efficiency are also improved than the onebyone insertion method.

A fast construction method for spatial index GBD-tree

This paper proposes a fast initial construction method of the GBD-tree. The GDB tree has proper characteristics for management of large amount of 2 or 3 dimensional data. However, the GBD-tree needs long initial construction time by originally proposed one-by-one insertion method. A fast insertion method has been proposed, but it needs large size of buffer capable to hold index information of all entries. The paper proposes another fast initial construction method. The method requires only limited size of work space (buffer). The experimental results show the initial construction time reduces into a third or a quarter of the one-by-one insertion method. The memory efficiency and retrieval efficiency are also improved than the one-by-one insertion method.