Database Reference
In-Depth Information
Figure 3-20.
Tables to show user purchase intent, aka recommendations
The relational model could simply provide User's friends that purchased certain Products, but to go deeper in the
recommendation it would be helpful to connect the users to friends who are nearby, share the same interests as well
as only show products that have the same interests, AKA “tags”. Although doing this in a relational model is certainly
achievable, the number of joins could impact performance as the network of users, products, locations and interests
begins to grow. In addition, the query plan would need to be known ahead of time or dynamically generated.
We can see that in the graph model, shown in Figure
3-21
, the has simpler way to display the interconnectedness
of each of the other four graph types as well as the ability to quickly connect intent with location. In addition to
creating an easy way to view, the query plan would not need to be precisely known ahead of time.
