The betweenness centrality of a node in a network measures how often that node appears on the shortest path between other pairs of nodes. Nodes that show up on many shortest paths score high, while those that rarely lie between others score low. Researchers use this idea widely in fields like social network analysis, transportation, and biology to identify influential or central players. This paper focuses on which nodes achieve the highest or lowest betweenness scores in a graph, calling those sets the betweenness center and betweenness periphery, respectively, and asks where these special sets tend to live within a graph's structure.
A key structural concept the authors use is that of a "block," which is a maximal well-connected piece of a graph that cannot be disconnected by removing a single node. The paper establishes conditions, based on simple properties like the diameter of the graph or the sizes of its blocks, that guarantee the most central nodes all cluster together inside one block rather than being scattered around. The authors also prove a surprisingly general result: any graph whatsoever can appear as the betweenness center of some larger graph, and the same is true for the betweenness periphery. This means there is essentially no restriction on what shape the set of most or least central nodes can take.
For trees, which are graphs with no cycles, the authors show through a new proof that the betweenness center always lies along a single path, a clean and tidy result. They also carry out a computational enumeration of all trees with up to 20 nodes, tracking how large the betweenness center can be relative to the whole tree. This gives a concrete quantitative picture of how central nodes are distributed in tree-structured networks and provides a resource for future researchers studying centrality in simpler graph families.