#1
  1. Traffic drop sleuth. :)
    SEO Chat Hero (2000 - 2499 posts)

    Join Date
    Jan 2010
    Posts
    2,065
    Rep Power
    1764

    How Google Determines Relevancy - In the Plex


    Here's more info that I'm learning from Steven Levy's book, In the Plex. I read last night about how Google learned to determine relevancy. I always wondered how (or even *if*) they knew that a search on "Moving to Phoenix" was relevant to "Arizona Real Estate". Or, how a search on "Gregory House" was not relevant to "Buy a home".

    So, according to the book, Google learns from its failures. If a user types in "info on dogs" they get a page of search results. Let's say that the results don't quite match what they are looking for, so they refine their search to say, "Info on puppies". This tells Google that "dogs" are relevant to "puppies".

    It then goes on to tell of how it had difficulties with some queries. For example, using the technique above Google quickly learned that "hot" was relevant to "boiling". But, if someone did a search for "hot dogs" they weren't likely to find pages on "boiling puppies". Instead, they tweaked the algorithm so that it figured out that "hot dogs" were relevant to pages about bread and processed meat and ball parks and barbecues. The algorithm is always changing in order to bring the user the most relevant pages to their search.


    They had an interesting section on how Google learned to determine whether someone was searching for a person. Google bought the rights to search through the entire volume of white pages. So, it learned what words were likely to be names. It then went through a learning process to determine whether someone was searching for a person, a location, or something else.

    I am blown away by how powerful Google is!

    And hey, I just realized this was my 1000th post on SEOChat! I love this place!

    Comments on this post

    • KernelPanic agrees : Grats on 1k posts!
    • EGOL agrees : congratulations on #1000!
    • distinctseo agrees : 1000!
    Last edited by Dr.Marie; Apr 15th, 2011 at 07:47 AM.
  2. #2
  3. Contributing User
    SEO Chat Adventurer (500 - 999 posts)

    Join Date
    Jan 2011
    Posts
    827
    Rep Power
    413
    You can also help Google understand if you're a person by using microformats. If you haven't read about those, look into it!
    Interested in SEO? Follow me: @cstechjoel or visit I'm The Best Looking Man In The World (According To Google!).
  4. #3
  5. Online Business Builder
    SEO Chat Adventurer (500 - 999 posts)

    Join Date
    Dec 2010
    Location
    India
    Posts
    697
    Rep Power
    369
    Originally Posted by Dr.Marie
    Here's more info that I'm learning from Steven Levy's book, In the Plex. I read last night about how Google learned to determine relevancy. I always wondered how (or even *if*) they knew that a search on "Moving to Phoenix" was relevant to "Arizona Real Estate". Or, how a search on "Gregory House" was not relevant to "Buy a home".

    So, according to the book, Google learns from its failures. If a user types in "info on dogs" they get a page of search results. Let's say that the results don't quite match what they are looking for, so they refine their search to say, "Info on puppies". This tells Google that "dogs" are relevant to "puppies".

    It then goes on to tell of how it had difficulties with some queries. For example, using the technique above Google quickly learned that "hot" was relevant to "boiling". But, if someone did a search for "hot dogs" they weren't likely to find pages on "boiling puppies". Instead, they tweaked the algorithm so that it figured out that "hot dogs" were relevant to pages about bread and processed meat and ball parks and barbecues. The algorithm is always changing in order to bring the user the most relevant pages to their search.


    They had an interesting section on how Google learned to determine whether someone was searching for a person. Google bought the rights to search through the entire volume of white pages. So, it learned what words were likely to be names. It then went through a learning process to determine whether someone was searching for a person, a location, or something else.

    I am blown away by how powerful Google is!

    And hey, I just realized this was my 1000th post on SEOChat! I love this place!
    we, as a SEO also make the Google job quite easy... there are few tags like <abbr>, <vcard> (address code) can also help Google to understand some confusing word (WI for Wisconsin) . also when we post something on the web we tag (include keywords in content, links, bold) certain page. i.e if you post for "hot dog" you'll surely use other keywords which are related to "hot dogs" and that includes hot dog roller grill, hot dog recipe, hot dog ingredients, blah blah... so what will Google do is generate a group of similar keywords and might deliver result based on that group. I thought this because of this video http://www.youtube.com/watch?v=hg8xgoULIIE though it is for Google news search engine.

    BTW... Congregates for your thousands post.
    Last edited by chiragparekh; Apr 15th, 2011 at 01:45 PM. Reason: Congregates for your thousands post.
  6. #4
  7. No Profile Picture
    Contributing User
    SEO Chat Explorer (0 - 99 posts)

    Join Date
    May 2010
    Location
    Montreal
    Posts
    50
    Rep Power
    15
    I bought the book today Miss Marie and it's your fault!

    Comments on this post

    • Dr.Marie agrees : Enjoy it! It's really interesting!
  8. #5
  9. No Profile Picture
    Contributing User
    SEO Chat Explorer (0 - 99 posts)

    Join Date
    Jan 2011
    Location
    Singapore
    Posts
    76
    Rep Power
    59
    I just googled 'boiled puppies'. lol
  10. #6
  11. Traffic drop sleuth. :)
    SEO Chat Hero (2000 - 2499 posts)

    Join Date
    Jan 2010
    Posts
    2,065
    Rep Power
    1764
    LOL! Now, you made me look it up too.
  12. #7
  13. Contributing User
    SEO Chat Discoverer (100 - 499 posts)

    Join Date
    Oct 2010
    Location
    New Zealand
    Posts
    422
    Rep Power
    227
    Just got my copy too

Similar Threads

  1. Ranking high on chat keywords
    By southpole854 in forum Public Site Reviews
    Replies: 5
    Last Post: Mar 12th, 2008, 12:55 PM
  2. When Google got it wrong
    By Alex324 in forum Google Optimization
    Replies: 10
    Last Post: Mar 12th, 2006, 03:10 AM
  3. Googleguy speaks! Some good clarifications
    By thewormman in forum Google Optimization
    Replies: 30
    Last Post: Jun 4th, 2005, 02:01 AM
  4. Article: new Google feature called Google Suggest
    By dirtdog1960 in forum SEO Chat Articles
    Replies: 0
    Last Post: Dec 12th, 2004, 01:09 AM
  5. Google - Tearing me up again..
    By clasione in forum Search Engine Optimization
    Replies: 2
    Last Post: Oct 7th, 2004, 04:00 PM

IMN logo majestic logo threadwatch logo seochat tools logo