PageRank

PageRank is an algorithm that assigns each node in a directed graph a score reflecting its importance, based on the idea that a link from an important node should count for more than a link from an unimportant one. It was originally developed to rank web pages by their incoming hyperlinks, but the same math applies to any directed graph, from citation networks to social graphs.

The algorithm works iteratively: each node starts with an equal score, then repeatedly redistributes its score evenly across its outbound links, with a damping factor that models a random walker occasionally jumping to an arbitrary node instead of following a link, so the process converges rather than concentrating all weight on whichever node happens to have the most incoming links. This makes it a specific instance of the broader family of eigenvector-based centrality measures, distinguished by its damping term and its origin in graph analytics for the web.

PageRank matters historically as the algorithm behind early Google search ranking, and it remains a component, alongside many others, in how search engine results page rankings and metrics like domain authority are computed today. A common misreading is treating PageRank, or a vendor's PageRank-like score, as a direct or complete measure of current search ranking; modern ranking systems combine hundreds of signals, of which link-based importance, and link prediction for missing connections, is only one part.