Dynamic Network Analysis (DNA), is an emergent field within the larger area of network analysis. At its simplest level, DNA adds a time element to the usual network structure, facilitating temporal analysis of the data.
There are many potential variables introduced when we add a time element to the network. Relationships between network nodes may strengthen, weaken, or even disappear as time unfolds. We may also witness physical movements in the network, the entry of new members, or the removal of existing nodes for a variety of reasons. In short, the network becomes increasingly complex.
What we can deduce from the preceding definition is that dynamic networks afford the possibility for greater exploration compared to traditional networks. We can see how networks are likely to evolve, witness the appearance and disappearance of links, and understand how a network changes (or is likely to change) over time. Here's what we'll cover in this chapter:
- When to use DNA
- The process of preparing data for DNA analysis in Gephi
- How to implement and view graphs within Gephi
- How to create GEXF files outside of Gephi
Let's begin by discussing when we should use dynamic network analysis.
There are a host of potential applications for DNA, including the following:
- Social media analysis, where friends and contacts are frequently changing
- Communication networks, such as corporate e-mail systems, where evolving patterns emerge over the course of days and weeks
- Political networks that change over time as entities gain or lose power
- Terrorist cells with frequent changes in structure driven by increasing membership and evolving network connections
- Disease modeling, where contagion rates can force rapid changes in the status of nodes within a network
In short, any network with relatively frequent changes over time will be a good candidate for DNA. Networks with infrequent or very slow changes (perhaps tenured faculty at a university or power grid infrastructure networks, to name just two examples) are often adequately addressed by static networks, as temporal analysis adds complexity while shedding little additional insight into network behavior.
There are two distinct types of DNA that can be created, as described here:
- A dynamic topological network, where member nodes can change positions, and appear or disappear at specific time intervals. This approach can be used to observe network growth achieved via new entrants, and to witness changes in structure due to movements within the network. If your wish is to see changes to an overall network, this is probably the direction to pursue.
- A network with dynamic attributes is different, with the focus being on changes to the nodes themselves, rather than the structure of the network. In this case we might observe the degree of growth of specific nodes across multiple time intervals. This approach is somewhat more challenging to implement, as it involves repeating each node at multiple stages rather than the customary single instance, which will require your source data to have a somewhat different structure. We'll take a look at how to do this later in the chapter.
It is important to note that the two types highlighted above are not mutually exclusive. They can be combined to detail the evolution of complex network behaviors where nodes and relationships emerge and vanish, but also change in stature over time.
Our first look will be focused on the processes needed to create common topology-based examples of dynamic network analysis. Later in the chapter, we'll take a similar walk to develop and create projects for attribute-based DNA.