Search Engine Optimisation

Search Engine Marketing

Web Site Development

 

Google patent summary

Techniques for finding related document [US 6,754,873 B1]

The methods documented in this Google patent describes techniques for identify and ranking related documents using link based analysis.

The Google patent describes how a system like Google can identify related documents by analyzing the backlinks of a document to find a set of index documents. The forward links from the index pages are then analyzed to find a related page set; the link structure of related pages is then analyzed to score and further refine the related page set. Pages are finally identified as been related when pages are linked by link structure and scored similarly.

The score of a page is derived from inbound link numbers from the index set. The ranking system employs a similar score reduction system as Google PageRank, with multiple inbound links from a single host reducing the value of inbound links from that host and links from index pages with a high volume of links will further reduce the score of each link.

Pages with a high volume of related web pages may be assigned as popular web pages. Pages assigned as popular will only be displayed as related to another page by Google if the relationship is symmetrical. i.e. if you have a page which has an inbound link from a page which also has a link to a popular site like Yahoo, and scores attained for both pages are similar when the backlink scores are calculated there would be an association made; however Google disassociates popular pages unless the site has a symmetrical relationship which can only be achieved if the scores for your site are similar when backlinks to yahoo are calculated.

When ranking the relationship between web pages, a preference is described which favors symmetrical relationships. i.e. page A has a list of related web pages which includes page B, page be also has a related page list which page A is included.

Conclusion

To view the results of this Google patent, input 'related:www.domainname.domain' into Google.

Any one can see that this functionality is not highly valuable to a user and has only limited use to a webmaster. Therefore inorder for it to be worthy of development by Google it must be included in the main algorithm used by Google to rank search result. It seems logical that this development is a refinement of the Google PageRank algorithm, which implements a limited score for sites which are related.