struc2vec: Learning Node Representations from Structural Identity
This addresses the challenge of representing structural roles in networks for applications like node classification, offering a novel approach beyond existing methods.
The paper tackles the problem of learning node representations that capture structural identity in networks, where nodes are identified by their structural roles rather than proximity, and presents struc2vec, a framework that outperforms state-of-the-art methods in capturing structural identity and improves classification performance in tasks dependent on it.
Structural identity is a concept of symmetry in which network nodes are identified according to the network structure and their relationship to other nodes. Structural identity has been studied in theory and practice over the past decades, but only recently has it been addressed with representational learning techniques. This work presents struc2vec, a novel and flexible framework for learning latent representations for the structural identity of nodes. struc2vec uses a hierarchy to measure node similarity at different scales, and constructs a multilayer graph to encode structural similarities and generate structural context for nodes. Numerical experiments indicate that state-of-the-art techniques for learning node representations fail in capturing stronger notions of structural identity, while struc2vec exhibits much superior performance in this task, as it overcomes limitations of prior approaches. As a consequence, numerical experiments indicate that struc2vec improves performance on classification tasks that depend more on structural identity.