Network visualization: why and what to consider?

Networks, nodes connected by lines, have become a symbol of complexity, for example as illustrations of AI and deep learning. A lot of big data is visualized and analyzed with networks as this is quick and intuitive, but how much can you learn from a network visualization? Which challenges are there in creating or understanding network images?

Visualization of a simple directed network

Why network visualizations?

A lot of systems can be modelled using networks. In examples such as social networks, or the road network, the network structure is obvious. But even if you consider a list of apparently independent people or objects, you can turn it into a network by computing similarities or some kind of distances between them. Network structures pop up everywhere and your need to understand networks will be ubiquitous.

Network visualizations are intuitive and insightful tools to understand networks as the human brain is made for quickly finding patterns in visual information. Images and visualizations can be very motivating. Images can convey an idea immediately. For instance, you can see at once whether the points all cluster together or whether they form several distinct groups.

Suitable network visualizations quickly reveal communities, or highly connected nodes (called hubs). If there are different groups, the network…