AN EFFICIENT ALGORITHM FOR COMPUTING PARTIAL TRANSITIVE CLOUSERES

Abstract

This paper foucses on the computation of partial transitive closures (PTC) in deductive databases. We present an algorithm that takes advantage of topological sort and a tagging technique to achieve better performance. The algorithm has two phases. The first phase, like many others, is to generate a topo logically sorted subgraph, containing only the reachable nodes and edges of the query. In the second phase, we evluate the nodes from sources to leaves in topological order. Since the reachable subgraph is already sorted, the evaluation of nodes can be fully sequenced. A tag is associated with each node traversed, indicating from which source nodes the node can be reached. This tagging technique simplifies the storage of intermediate results and speeds up the computation. The new algorithm is simple, easy to implement, and yet efficient. It compares favorably to well-known algorithms, expecially when the unmber of source nodes is not large.