Detecting Communities using Absorbing Markov Chains

Community detection is an important part of network analysis. The goal is simple: to detect how nodes in the graph ought to be grouped into communities. Some algorithms wholly partition the nodes; others allow for some nodes to be considered ”community-less”. Community detection has several applications: for example, suggesting connections (as in, say, Facebook) and determining network structure. As we’ll see in this paper, community detection can also be quite useful to derive features for classification tasks where the data includes interactions that are not easily translatable into features.