1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Google, Content and everything in between: Explained!

Discussion in 'Black Hat SEO' started by satyawrat, Mar 4, 2011.

  1. satyawrat

    satyawrat Jr. VIP Jr. VIP

    Joined:
    Jul 8, 2009
    Messages:
    919
    Likes Received:
    1,181
    Occupation:
    Hustler
    Location:
    Gurgaon
    Home Page:
    SEO20 recently made a really good thread explaining how spinning alone is not going to work. I am going to go a little deeper and explain how Google works out relation between keywords, content and niche.

    The first step is to clearly understand what algorithm is:
    To keep things simple just think of it as set of instruction given to computer to process data.
    Now its time for the big statement:
    Google did NOT change algorithms! Its false!
    What it did is it gave more weightage to the algorithm which already existed and maybe fine tuned a little.

    What algorithms are we talking about here?
    There are not many options available as of now to process natural language. The best option is Singular Value Decomposition and Latent Semantic Analysis. I will be talking about them in detail from the SEO point of view.

    Let us first talk about LSA:
    Google has its own LSA algorithm, but it is not far from the one already patented by a group of people. You can find more info on it here:
    Code:
    http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PALL&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.htm&r=1&f=G&l=50&s1=4839853.PN.&OS=PN/4839853&RS=PN/4839853


    LSA Explained:
    Consider a matrix, the rows correspond to terms/keywords and the columns correspond to documents/niche/categories.
    but such a matrix will be really huge, so they use low-rank approximation in order to eliminate some "noise" and make the process faster. If you want to know more about low-rank approximation, let me know i will make a post on it.
    --------------------

    The second method is a cross between Singular Value Decomposition and vector space model. Its going to take considerable effort to sum it up for you, if you are not bored by this mathematical post then let me know i will explain them in detail too.


     
    • Thanks Thanks x 3
  2. JohnnyRabbit

    JohnnyRabbit Supreme Member

    Joined:
    Sep 29, 2010
    Messages:
    1,228
    Likes Received:
    299
    Informative.
     
  3. satyawrat

    satyawrat Jr. VIP Jr. VIP

    Joined:
    Jul 8, 2009
    Messages:
    919
    Likes Received:
    1,181
    Occupation:
    Hustler
    Location:
    Gurgaon
    Home Page:
    The best is yet to come :p

    LSA Continued:
    COnsider a matrix. In this matrix, the we will pay attention to two words more closely: "rose" and "lamborghini".
    Now remember the rows correspond to the keywords and columns correspond to the documnets/niche. If "rose" has higher frequency of repetitions in the document "flowers" the algorithm will deem the word "rose" to be related to "flower" and the word "lamborghini" to be related to "cars".

    Another important point: if you are trying to rank for flower, you not only need "flower" as the keyword but you would also have to use other words which the algorithm might find similar like "rose"

    I am tired now, I will be making more posts. Hit the thanks button so I know you are liking it!
     
    • Thanks Thanks x 6
  4. donArmando

    donArmando Regular Member

    Joined:
    Oct 28, 2010
    Messages:
    407
    Likes Received:
    256
    of course you have to think about what algorithm might find similar
    example : cars , bmw , wolswagen etc because all point us to cars and all google sees as Cars.
     
  5. cbnoob

    cbnoob Senior Member

    Joined:
    Sep 27, 2010
    Messages:
    967
    Likes Received:
    455
    Don't know I like your text or not but really like to understand it