Cardinality Estimation using Label Probability Propagation for Subgraph Matching in Property Graph Databases

Abstract

Estimating query result cardinality is a central task of cost-based database query optimizers, enabling them to identify and avoid excessively large intermediate results. While cardinality estimation has been studied extensively in relational databases, research in the setting of graph databases has been more limited. In this paper, we address the problem of cardinality estimation for subgraph matching on property graph databases. Our novel cardinality estimation technique starts from a small amount of statistical information about node labels and relationship types, which is propagated along the graph query pattern in terms of label probabilities. Additionally, estimation quality can be improved by providing information about labels or properties to our technique, if available. In our experimental evaluation, we compare our approach to state-of-the-art cardinality estimation techniques for subgraph matching for property graph, RDF, and relational databases, and we demonstrate that our technique offers the best trade-off between accuracy and efficiency.