Research And Improvement Based On Bitemporal Index 4R-tree

4R-tree index is a practical bitemporal index, which consists of four R-trees named R1, R2, R3 and R4 respectively. It can handle time-variable Now and UC efficiently, and can be employed in any Database Management System that supports R-tree.Tree R2 indexes data with non-fixed transaction-time intervals by physically storing only vertical lines, but owing to the disadvantage that R-tree can not index lines efficiently, the query performance in tree R2 is too bad, this may have the reverse effect on the performance of 4R-tree index. Hence, this paper provides a solution to reduce the query costs in tree R2. This is achieved by transforming the minimum bounding rectangles of the index items in non-leaf nodes and the data items in leaf nodes from two dimensions, that is valid-time dimension and transaction-time dimension, to three dimensions, that is valid-start-time dimension, valid-end-time dimension and transaction-time dimension. So that, this solution includes the transformation of the two-dimensional intervals from the original tree R2 into the three-dimensional points and the accompanying query transformation. The modified tree R2 takes advantage of its efficient point-search in three-dimensional space instead of interval-search in two-dimensional space.Because each R-tree in 4R-tree index is unattached to others, and only tree R2 is modified, the modified 4R-tree index is directly tested in two aspects of update costs and query costs to demonstrate the performance of the modified tree R2 in the end of this paper. Concluded from the experimental results, the total update costs of 4R-tree index increase a little, but the query costs reduce more. At the same time, this conclusion also reflects that the query efficiency in the modified tree R2 is improved.