Clustering Coefficient A clustering coefficient is how likely a friend knows another friend, or a connected node is connected to another connected node. Social networks or sites like Facebook have a high clustering coefficient. In the case of a social network someone’s friend is more likely to know a friend than someone completely random in the world. This by itself is not enough to create small world networks, networks where there is a low amount of links needed to connect any 2 nodes. Clusters of people such as a club or a social circle need to be connected to other clusters. These connections can be made by different affiliations, for example in the above image the top right cluster connects Alan, Emil, and myself. We all know each other through work. Even though Nik a game developer seems very distant from Emil, Emil knows me who studied game design under Nik. Clustering can by high like the cluster in the bottom right, but there needs to be seemingly random links to other clusters as illustrated in the center of the image. Websites increase their clustering coefficient with a few different methods. If I interact with Marco on Facebook, Caroline or Barb might see that in their newsfeed and there might be a chance either of them would friend Marco.Tumblr attempts to increase clustering by adding a list of other accounts you might like in their sidebar, assuming these suggested accounts are chosen based on who you follow, follow’s. This is not the same as affiliated connections, which is which is better represented as the radar, which I’ll explain in my Affiliated vs Clustered post next week. If your looking for a solid number for clustering on a network you would divide the average amount of possible node links by the average amount of actual node links. If you kept track of this on a social website then you can make adjustments to your site layout and map to know if they’re helping you replicate a small world network.A lot of what I share was learned through a few books, but for this post mostly Six Degrees by Duncan Watts.

Clustering Coefficient

A clustering coefficient is how likely a friend knows another friend, or a connected node is connected to another connected node.

Social networks or sites like Facebook have a high clustering coefficient. In the case of a social network someone’s friend is more likely to know a friend than someone completely random in the world. This by itself is not enough to create small world networks, networks where there is a low amount of links needed to connect any 2 nodes. Clusters of people such as a club or a social circle need to be connected to other clusters. These connections can be made by different affiliations, for example in the above image the top right cluster connects Alan, Emil, and myself. We all know each other through work. Even though Nik a game developer seems very distant from Emil, Emil knows me who studied game design under Nik. Clustering can by high like the cluster in the bottom right, but there needs to be seemingly random links to other clusters as illustrated in the center of the image.

Websites increase their clustering coefficient with a few different methods. If I interact with Marco on Facebook, Caroline or Barb might see that in their newsfeed and there might be a chance either of them would friend Marco.
Tumblr attempts to increase clustering by adding a list of other accounts you might like in their sidebar, assuming these suggested accounts are chosen based on who you follow, follow’s.
This is not the same as affiliated connections, which is which is better represented as the radar, which I’ll explain in my Affiliated vs Clustered post next week.
If your looking for a solid number for clustering on a network you would divide the average amount of possible node links by the average amount of actual node links. If you kept track of this on a social website then you can make adjustments to your site layout and map to know if they’re helping you replicate a small world network.

A lot of what I share was learned through a few books, but for this post mostly Six Degrees by Duncan Watts.

Tagged: social media, .