Update: Orion has provided a link to the duplicated content patents presentation.
Dr. E. Garcia, of Mi Islita.com, also known as "Orion"?? on the Search Engine Watch forum presents the next portion of the Search Algorithms session moderated by Search Engine Watch editor, Chris Sherman.
Dr. Garcia’s talk centered on the 2003 Google Patent, “Detecting Query Specific Duplicate Documents”.
The 2003 patent covers how Google might improve quality of search results by comparing snippets. He also points out why paying attention to the snippets of text that Google presents in search results can improve click through rates.
Highlights of the patent:
– User performs a search
– Candidate results are ranked by relevancy
– Query specific filters are applied
– Final set of SERPs (search engine results pages) are shown to user
With each search result from the main index, Google generates query relevant snippets from SERPS.
Snippets are linear, 15 term or 100 character sliding windows based on the location of the matching keyword query.
Google will count the term frequency within the window and then compare a current snippet with snippets already in the final set.
Dr. Garcia explained a formula for how Google might calculate whether to show one snippet over another and how information form a snippet can affect ranking.
The value in snippet optimization for overall search engine optimization and web site copywriting is to pay attention to the location of target keywords within the body text of a document as well as other occurrences of related phrases within a similar proximity. More effective snippets can increase click throughs on natural search results.