Connections are what define a graph. Without links, nodes are just a table of data. While most of this book has discussed links as only one or two links between nodes, oftentimes there are multiple links. For many objectives, these multiple links may be aggregated into a singular link. However, for some types of analyses and applications, you want to keep those many links and then have approaches to view, filter, and separate different subsets of the graph based on these links. Applications where it is important to find and identify a few anomalies in the data (such as fraud detection or cybersecurity) are examples where it is important to retain the individual links.
At very simple level, you have undirected links. You can perform a lot of graph analysis at this level, and most of the examples in the book up to this point utilize undirected links. You have also seen a few examples with directed links.
But in the real world, relationships can be much more complex than simple directed and undirected links. For example, let’s say that you are a user of LinkedIn. You can query a particular person, and if you are directly connected, the relationship is immediately shown. LinkedIn will also show all the types of connections between you and the other person (for example, field of study, skills and expertise, location, school, group, and so on). Figure 9-1 shows the many links between the two authors of this book.