Applying k-vertex Cardinality Constraints on a Neo4j Graph Database

As with any other database solution, graph databases also need to be able to implement business rules related to a given application domain. At the moment, aside from integrity constraints, there is a limited number of mechanisms for business rules implementation in Graph Database Management Systems (GDBMSs). The underlying property graph data model does not include any formal notation on how to represent different constraints. Specifically, this paper discusses the problem of representing cardinality constraints in graph databases. We introduce the novel concept of k-vertex cardinality constraints, which enable us to specify the minimum and maximum number of edges between a vertex and a subgraph. We also propose an approach, which includes the representation of cardinality constraints through the property graph data model, and demonstrate its implementation through a series of stored procedures in Neo4j GDBMS. The proposed approach is then evaluated by performing experiments on synthetic and real datasets to test the influence of checking cardinality constraints on query execution times (QETs) when adding new edges. Additionally, a comparison is performed on synthetic datasets with varying outgoing vertex degrees in order to gain an insight into how increasing the vertex degree affects QETs. In general, the results obtained for each test scenario show that the implemented k-vertex cardinality constraints model does not significantly affect QETs. Also, the results indicate that the model is dependent on the order of the underlying k-vertex cardinality constraints and outgoing vertex degree in the dataset.