ABSTRACT
Throughout my career as a data scientist, I have always been dubious about the utility of network diagrams for the understanding of unstructured data. e problem has always been that, while it is very easy to draw nodes connected by arcs and display them in a stunning visualization on a computer screen, it is very oen unclear what real insight is gained from the exercise. e fact that a picture looks compelling does not in fact make it useful. e reason for this is oen that the nodes and arcs have not been suciently thought through so that what is being connected in the graph has some relevance to a scientic problem. But even if the nodes are the important scientic entities and the edges represent some signicant connection between them, the mere fact of putting the nodes and edges onto a screen as a graph is oen not useful in and of itself.
