The 'Reasonable Surfer' patent

15 December 2011

In 2010 Google was granted a patent called "Ranking documents based on user behavior and/or feature data" which it had originally applied for in 2004.

The patent describes how when a search engine spider crawls through and indexes web pages, it creates a model which is used to rank those pages.  This process looks at features associated with web pages which contain links, but also the destination page that the link points to and also the link and linktext  itself.

The patent also covers user/visitor behaviour in the way website visitors use and navigate through a website, the links they use to do so and the software (e.g. browser plugins and toolbars) that may be used to gather such information.

The following three lists summarise the type of data that the new patent describes.

Features associated with links

• Font size
• Position on the page
• Position in lists on the page
• Font colour
• Anchor text
• 'Topical Clusters' associated with the anchor text
• Words preceding and following the link.
• Link destination

Features associated with the web page on which the link is found

• The URL
• Number links on page
• Other words on the page
• 'Topical clusters' associated with the page
• The association between the Topical Clusters associated with the source page and the link

User behaviour

• Interaction with the content on the page
• User language
• Users interests
• Search queries used
• The frequency of link use
• The frequency of times no links are used and bounce rates

You can read an excellent summary of the 'Reasonable Surfer' patent on the SEO by the sea website.