Modeling - Practical Neo4j - page 22

Figure 3-3. A ER model with attributes
Challenges in Using Entity-Relationship Modeling with Neo4j
Traditional entity-relationship models accept information and content that can be freely and easily contained within
a relational database and are typically only a good match for a relational structure. In fact, they are insufficient for
models in which the data cannot be suitably represented in relational form, as is the case with frequently changing,
semi-structured data. One of the biggest challenges for many applications is the possible frequency and scope of change
to the way model is structured. As detailed in Chapter 1, these types of modifications for relational systems are nontrivial,
involve at least moderate risk, and are often significant causes for changes from one database platform to another.
Modeling with Neo4j
This section begins to build out the model for the application to be discussed in the later chapters of the topic. The
model contains some likely familiar themes in terms of its structure and includes five areas that have been identified
as the most significant portions of both consumer and business data: social , intent , consumption , interest , and location
graphs. These five graph types are certainly not the only use cases that make sense for Neo4j, but they are in wide use
and intrinsically shaped.
As part of our examination of the graph model for these areas, we will examine the companion model structure
as designed for a relational database. As noted in the data model overview section, a divergence takes place when
modeling relationships within a relational database versus modeling within Neo4j. The divergence is not significant
in terms of the data being captured, but, as Table 3-1 shows, the main components of an entity-relationship model in
Neo4j may be known by different names and take vastly different shapes.