Modeling - Practical Neo4j - page 26

Figure 3-9. Graph diagram with mutual connection. The direction implies who made the request
Figure 3-10. Graph diagram with specific directed connections
While deciding the manner in which your social model should be established, it is important to consider that
there is more than just a technology decision at stake, but, potentially, a business decision as well. While both models
allow for exploring connections in either direction from a technical standpoint, the bidirectional relationship implies
that only one user action needs to occur in order to establish a mutual connection.
In addition, using the bidirectional or mutual option, by definition, will reduce the number of relationships
comparatively by 50 percent. The problem of dense nodes—think of any celebrity who might have millions of
followers but only follows a few other users—is less a factor in performance in the latest version of Neo4j. However,
directional relationships can sometimes have an impact and need to be considered carefully. For the purposes of the
book's example application, we will consider the directional relationship for the social aspect, such as the connection
method found in applications such as Twitter.
Interest Graph
The interest graph is closely connected to the intent graph. However, the interest graph is principally concerned with
the connecting a person with her specific interests. In that sense, the interest graph would allow for an application to
make recommendations regarding related items of interest much in the same way a thesaurus can offer synonyms
of a specific word. When combining the interest graph with a person's demographic or social graph, an application
can make recommendations that typically have a higher degree of connectedness and relevance. Figure 3-11
demonstrates how an interest graph could be created within a relational model.