Ranking and clustering of nodes in networks with smart teleportation

TL;DR

This paper introduces a novel approach to random walk-based ranking and clustering in directed networks by using link-based teleportation, which offers smoother trade-offs and more robust results compared to traditional node-based teleportation.

Contribution

It proposes link teleportation as an alternative to node teleportation, improving robustness and reducing teleportation bias in network ranking and clustering.

Findings

Link teleportation enables smoother trade-offs in ranking.

Not recording teleportation steps reduces teleportation effects.

Results show improved robustness in clustering.

Abstract

Random teleportation is a necessary evil for ranking and clustering directed networks based on random walks. Teleportation enables ergodic solutions, but the solutions must necessarily depend on the exact implementation and parametrization of the teleportation. For example, in the commonly used PageRank algorithm, the teleportation rate must trade off a heavily biased solution with a uniform solution. Here we show that teleportation to links rather than nodes enables a much smoother trade-off and effectively more robust results. We also show that, by not recording the teleportation steps of the random walker, we can further reduce the effect of teleportation with dramatic effects on clustering.