Cost-Based Query Optimization for Multi Reachability Joins

There is a need to efficiently identify reachabilities between different types of objects over a large data graph. A reachability join (R-join) serves as a primitive operator for such a purpose. Given two types, A and D, R-join finds all pairs of A and D that D-typed objects are reachable from some A-typed objects. In this paper, we focus on processing multi reachability joins (R-joins). In the literature, the up-to-date approach extended the well-known twig-stack join algorithm, to be applicable on directed acyclic graphs (DAGs). The efficiency of such an approach is affected by the density of large DAGs. In this paper, we present algorithms to optimize R-joins using a dynamic programming based on the estimated costs associated with R-join. Our algorithm is not affected by the density of graphs. We conducted extensive performance studies, and report our findings in our performance studies.