Set Containment Joins: The Good, The Bad and The Ugly

Abstract

Efficient support for set-valued attributes is likely to grow in importance as object-relational database systems, which either support set-valued attributes or propose to do so soon, begin to replace their purely relational predecessors. One of the most interesting and challenging operations on set-valued attributes is the set containment join, because it provides a concise and elegant way to express otherwise complex queries. Unfortunately, evaluating these joins is difficult, and naive approaches lead to algorithms that are very expensive. In this paper, we develop a new partition based algorithm for set containment joins: the Partitioning Set Join Algorithm (PSJ), which uses a replicating multi-level partitioning scheme based on a combination of set elements and signatures. We present a detailed performance study with a complete implementation in the Paradise object-relational database system. Our results show that PSJ outperforms previously proposed set join algorithms over a wide range of data sets.