A Formal Design Framework for Practical Property Graph Schema Languages

Abstract

Graph databases are increasingly receiving attention from industry and academia, due in part to their flexibility; a schema is often not required. However, schemas can significantly benefit query optimization, data integrity, and documentation. There currently does not exist a formal framework which captures the design space of state-of-the-art schema solutions. We present a formal design framework for property graph schema languages based on first-order logic rules, which balances expressivity and practicality. We show how this framework can be adapted to integrate a core set of constraints common in conceptual data modeling methods. To demonstrate practical feasibility, this model is imple-mented using graph queries for modern graph database systems, which we evaluate through a controlled experiment. We find that validation time scales linearly with the size of the data, while only using unoptimized straightforward implementations.