Algorithms For Dummies

Implementing PageRank

Representing this chain of trust mathematically requires simultaneously deter-

mining how much authority your page has, as measured by inbound links, and 

how much it donates to other pages by outbound links. You can achieve such com-

putations in two ways:

---

CHAPTER 11

  Getting the Right Web page 

     213

 

»

Simulation: Uses the behavior of a web surfer who browses randomly on the 

web (a random surfer). This approach requires that you recreate the web 

structure and run the simulation.

 

»

Matrix computation: Replicates the behavior of a random surfer using a 

sparse matrix (a matrix in which most data is zero) replicating the web 

structure. This approach requires some matrix operations, as explained in 

Chapter 5, and a series of computations that reach a result by successive 

approximation.

Even though it’s more abstract, using the matrix computation for PageRank 

requires fewer programming instructions, and you can easily implement it using 

Python. (You can try the PageRank algorithm on real-world sites using an 

automatic PageRank checker, such as 

http://checkpagerank.net/index.php

. 

Unfortunately, the program may produce inaccurate results for newer sites 

because they haven’t been crawled properly yet, it can give you an idea of what 

PageRank is like in practice.)