A dynamic network is a graph evolving in time, where nodes and edges can be added, modified, or even deleted. The problem of community detection becomes much more complex because communities can do the following:
- Appear
- Grow
- Be reduced
- Fuse with another community
- Be split
- Disappear
A community can also stay unchanged or even temporarily vanish only to appear again sometime later. One technique to solve such problems consists of using snapshots of the graph at different times. A static algorithm such as the ones studied in this book can then be used on each of the snapshots. However, when comparing the communities discovered in two consecutive snapshots, it will be hard to decide whether the differences are due to the real community evolution or to the algorithm instability (think about the resolution limit of the Louvain algorithm). Many solutions have been proposed to solve this issue by using smoothing techniques. For instance, you can build an algorithm...
