The graph structure that most graph databases including Neo4j use inherently classifies them into property graphs. A property graph stores the data in the form of nodes. These nodes are linked with relationships that provide structure to the graph. Relationships must have a direction and a label and must exist between a start node and an end node (dangling relationships are not permitted). Both nodes and relationships can have properties, which are key-value pairs that store characteristic information about that entity. The keys have to be strings that describe the property, while the value can be of any type. In property graphs, the number and type of properties can vary across the entities. You can basically store all the metadata about that entity into it in the form of properties. They are also useful in applying constraints to the queries for faster access with the help of indexes.

A simple property graph

These are all that are required to create the most sophisticated and rich semantic models for the graphs. Diagrams are an excellent way to view and design graph structure models since they clarify the data model. However, in order to implement these models in the Neo4j graph systems, you will need tools such as Cypher that provide elegant constructs to design the most complex of systems, some of which we will see later in the chapter.