Abstract

We saw in the introduction that the node set of many networks can be partitioned into several groups based on the node attributes or on the node’s behaviour. For example, the members of the karate-club network split into two groups (Zachary, 1977), while the blogs of the political blog network are labelled as being either liberal or conservative. Community detection (also referred to as community recovery or graph clustering) consists in inferring the latent community structure based on the node’s interactions.