Patent

The original PageRank patent, assigned to Stanford and licensed exclusively to Google. Describes ranking nodes in a linked database by recursively weighting inbound links by the rank of the linking node.

The historical data patent, describing document inception dates, content update rates, link growth rates, anchor text change and domain registration signals as scoring inputs. Filed December 2003, published March 2008.

Part of Anna Patterson's phrase-based indexing family. Identifies meaningful phrases and uses co-occurrence — which phrases predict the presence of other phrases — to rank and personalise results. Filed July 2004, published August 2009.

Known in the industry as the reasonable surfer patent. Describes weighting links by the probability a user would click them, based on features such as position, font, anchor text and link context. Filed June 2004, issued May 2010.

Describes using aggregated usage statistics, such as how often documents are selected and how long users engage with them, as inputs to document retrieval and scoring.

Frequently cited in analyses of site-level quality scoring and of query-and-click based ranking modifiers. Filed September 2012, issued March 2014.