Skip to main content icon/video/no-internet

Core-Periphery Structure

A core-periphery structure is a type of network structure in which core actors are densely interconnected, whereas periphery actors are sparsely interconnected and only connected to core actors. An ideal core-periphery structure consists of a core where actors are all linked with each other and a periphery where actors are only linked to those in the core but no one in the periphery. While the ideal core-periphery structure is very rare in practice, many social networks, as well as other complex networks, resemble a core-periphery structure to varying degrees, with a densely interconnected core surrounded by a sparse periphery. This entry discusses key features of the core-periphery structure, introduces popular methods for detecting such structures, and presents some examples.

The defining feature of a core-periphery structure is a partition of a network into two sets, a well-connected core and a sparsely connected periphery. There has been no consensus about the exact extent of the connections between the core and the periphery, however. Some scholars suggest a core actor should be connected to all periphery actors, while others contend that it only needs to be connected to some periphery actors. The common ground is that core actors are expected to be reasonably well connected to periphery actors.

Usually only a relatively small number of actors are located in the core, but they play a prominent role in the overall network. Compared with periphery actors, they possess more power and enjoy more prestige due to their structural advantage. They are also more instrumental in diffusing information, innovations, and social norms. Dependency may emerge as periphery actors become dependent on the core for key resources embedded in the network.

Detecting Core-Periphery Structure

Core and periphery actors in a network play different roles. The detection of a core-periphery structure sheds light on both an actor’s structural position and the functioning of the overall network. Scholars have developed a variety of quantitative methods to detect and measure a core-periphery structure.

The most well-known detection method is the discrete approach. It is based on the comparison of the observed network with the ideal core-periphery network that consists of a fully connected core and a periphery that has no internal connections. A network exhibits a core-periphery structure if its correlation with the ideal core-periphery network is high. This discrete approach is essentially a binary classification process that partitions all actors into two categories. A limitation of this approach is its oversimplification as it merely identifies two classes of actors. In the core, however, some actors may be more core-like than others; similarly, periphery actors are not all peripheral to the same degree. The discrete approach cannot reveal those nuances within each category.

Another popular method is built on the continuous approach in which each actor of the network is assigned a quantitative coreness score. Some algorithms have been proposed to calculate coreness scores. These scores are on a continuous spectrum. Actors with greater coreness scores are located closer to the core, whereas those with lower scores are positioned more toward the periphery. In doing so, the continuous approach is better at capturing nuanced variations in the core/periphery status among actors.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Sage Recommends

We found other relevant content for you on other Sage platforms.

Loading