DM8 - Link Analysis

It assigns an importance score to a web page based on the sum of the importance scores of the pages that link to it, operating as a recursive voting system.

The vote is distributed equally among all its outgoing links;if a page has an importance of $r$ and an out-degree of $n$, each neighbor receives $r/n$ votes.

An iterative algorithm that repeatedly multiplies the transition matrix by the rank vector ($r^{t+1}=M{r}^t$) until the vector values converge to a steady state.

It represents the principal (dominant) eigenvector of the column-stochastic transition matrix, corresponding to the eigenvalue of $1$.

A node with incoming edges but no outgoing edges, causing the standard PageRank random walk to leak or lose probability mass over time.

A group of nodes with no links to the outside network, causing the standard random surfer to get stuck and artificially absorb all the PageRank mass.

It introduces a probability ($1-\alpha$) that the random surfer will jump to a completely random page in the entire network instead of following an outgoing link.

Typically around $0.85$ (or between $0.8$ and $0.9$).