An algorithm for tree-query membership of a distributed query

The aim is to process distributed queries ef ficiently. The cost of communications between sites is dominant in processing such queries. It is assumed that the amount of data transferred determines the transmission cost to a large extent. Thus, it is desirable to minimize the amount of transmitted data. Bernstein-and Chiu [2] classified queries into two types: tree and cyclic queries. They defined an operation called semi-join which requires minimal transfer of data between sites. Then they showed that tree queries can always be answered by semi-joins but cyclic queries may not. An algorithm to decide whether a query is cyclic or not was presented in their paper. Their algorithm works when the number of domains in common between any two relations is no more than one. The aim of this paper is to generalize their algorithm. Specifically, we present a conceptionally simple algorithm which decides the type of a query when the number of domains in common between two relations may exceed one. An implementation of the algorithm is outlined. The algorithm runs in 0(max(e,e')) time and O(e) space complexity where e and e' are the number of edges in the transitive closure of the join graph and the query graph respectively.