Personalized PageRank with Node-dependent Restart
This is an incremental theoretical extension for graph analysis and web ranking applications.
The paper tackles the problem of generalizing Personalized PageRank by introducing two variants with node-dependent restart probabilities, showing they coincide in the original constant case and derive a symmetry property connecting direct and reverse versions.
Personalized PageRank is an algorithm to classify the improtance of web pages on a user-dependent basis. We introduce two generalizations of Personalized PageRank with node-dependent restart. The first generalization is based on the proportion of visits to nodes before the restart, whereas the second generalization is based on the probability of visited node just before the restart. In the original case of constant restart probability, the two measures coincide. We discuss interesting particular cases of restart probabilities and restart distributions. We show that the both generalizations of Personalized PageRank have an elegant expression connecting the so-called direct and reverse Personalized PageRanks that yield a symmetry property of these Personalized PageRanks.