Two Dominance Searching Based Temporal Join Algorithms

Abstract

In this paper, we are concerned with the problem of efficient processing of temporal join operation relations. By mapping time intervals to points in the plane, we first show that the problem of determining the set of all matching tuples of a temporal join is equivalent to a dominance searching problem in the plane. Then, by using an efficient data structure for solving the dominance searching problem as an index for the inner relations, we propose a new nested-loops based temporal join algorithm. For the cases where the index for the inner relations is too large to fit in the primary memory, we propose anther partition_based temporal join algorithm which doesn't need any index for the operand relations. Finally, in order to provide more efficient processing of temporal join, we propose a cluster scheme and an index scheme to support efficient strong of tuples and direct access of matching tuples.