Inside Page Ranking

INS ID E P AG E R ANK

B ejo Thampi S 7 CS E R oll No:-14 S NG C E On :-26/08/2009 G uided B y : Mr. S ujith K umar

C O NTE NTS  Introduction  P ageR ank  P ageR ank

calculation  O ptimization of P ageR ank  R emoval of D angling P ages  C onclusion  R eferences

Introduction  R epresents

how important a page is on the

web.  Numeric

value between 0 and 10

 O ld

Algorithms cons idered only the content of the page causing web s pamming

 C ons iders

both content and anchor text (BackLinks)

P ageR ank  P ageR ank

is a link analys is algorithm.

 D etermines

a page’s ranking in the search

res ults . A

trademark of Google which is patented to S tanford Univers ity.

C ontinue…  Q uoting

from the original Google paper, P ageR ank is defined as: • a link from page A to page B as a vote, by page A, for P age B. • it also analyzes the page that casts the vote.

C alculation of P ageR ank  Two

M ethods

 S implified

Algorithm and D amping

Algorithm  Based

probability distribution

S implified Algorithm In

the general cas e, the P ageR ank value for any page u can be expressed as:

D amping Algorithm  P R (V)=

d

(1-d) + d(P R (V 2)/1)

(damping factor) =0.85

C alculated

using S imple Iterative M ethod

E xample P age A

E ach

P age B

page has one outgoing link

i.e. hA

= 1 and hB = 1

G uess1 S tart a

guess at 1.0 P R (A) = (1 – d) + d(P R (B )/1) P R (B ) = (1 – d) + d(P R (A)/1)

i.e.

P R (A) = 0.15 + 0.85 * 1 = 1 P R (B ) = 0.15 + 0.85 * 1 = 1

numbers aren’t changing. S o consider this as a good guess.

here

we

G uess2 guess at 0 ins tead and re-calculate: P R (A) = 0.15 + 0.85 * 0 = 0.15 P R (B ) = 0.15 + 0.85 * 0.15= 0.2775

And again: P R (A) = 0.15 + 0.85 * 0.2775= 0.385875 P R (B ) = 0.15 + 0.85 * 0.385875= 0.47799375

C ontinue… O n 39th iteration , P R (A)=.999999 and P R (B )=1.0000000

O n40th iteration, P R (A)=1.000 and P R (B )=1.000. And average P ageR ank =1.00000

P rinciple:- the “normalized probability dis tribution” (the average P ageR ank for all pages) will be 1.0.

Variations  Google •

 •

Toolbar R ank The Google Toolbar's P ageR ank feature dis plays a vis ited page's P ageR ank as a whole number between 0 and 10. S E R P R ank R es ult returned by search engine in res ponse to a keyword query.

O ptimization of P ageR ank 

Three fundamental areas to look at when trying to optimize the P ageR ank for site:

1.

The links you choose to have link to you, i.e., which ones you choos e, and how much effort you put in to getting them.

C ontinue… 2. Who you choose to link out to from your site  M aximising

P ageR ank Feedback and minimising P ageR ank leakage

3.The internal navigational structure and linkage of your pages  distribute

P ageR ank within your site.

Internal Linking Hierarchical

Courtsey: http://www-db.Stanford.edu/~backrub/google.html

C ontinue… Looping:-

Courtsey: http://www-db.Stanford.edu/~backrub/google.html

R emoval of D angling P ages  S imply

links that point to any page with no outgoing links.

 These

links are simple pages that not downloaded yet.

 R anking

 Need

is affected.

to be removed http://www.iprcom.com/papers/pagerank/index.html

C ontinue… E liminates

dangling pages before P ageR ank C alculation.

D one

by introducing a dummy page.

A link to itself and is pointed by every dangling page.



C ontinue…

Courtsey: http://www.iprcom.com/papers/pagerank/index.html

C onclusion  P arameter

involved in G oogle’s ranking of the answers to a given query.

 O ptimal computation  O ptimization

of P ageR ank

of P ageR ank

R eferences “The G oogle S tory” - D avid A Vise 2. htp://www.iprcom.com/papers/pagerank/index.html 3. http://en.wikipedia.org/wiki/P ageR ank 4. http://www.siteall.com/guide/ 1.

THANK YOU

Q UE S TIO NS ??