Temporal PageRank
https://link.springer.com/chapter/10.1007/978-3-319-46227-1_42
We call a graph dynamic when the number edges and nodes it has varies over time. Then we call a graph temporal if it is dynamic and we know each node and edge given a particular time. The authors of this study derived a natural generalization of PageRank to apply to temporal graphs, which they call Temporal PageRank. Since their model generalizes PageRank, when they consider a temporal graph that stays constant over time, Temporal PageRank converges to our PageRank. After experiments on three data sets where they applied this model, they concluded that the model functions in the case of Temporal Graphs.
I chose this particular paper because I questioned how well PageRank works in practice in class, since pages vary so greatly in real social networks such as Facebook and Twitter, two of the datasets they evaluated. Their model accounts for my confusion, and they mention that some work has been done with dynamic graphs that involves evaluating PageRank every time a node or an edge changes which has still worked but not as well.
I was also curious about the connection between PageRank and random walks after reading Chapter 14. They showcase the advantages of interpreting PageRank in terms of random walks by using temporal random walks, a generalization of random walks to temporal graphs, to define their new PageRank. Their derivation of the math behind the model follows from results in linear algebra for random walks by observing the qualities of their adjacency matrices much like the math behind the original PageRank.