Brain Dump

Page Rank

Tags
text-processing

Googles search ranking algorithm.

Its [see page 24, based] on a simple intuition:

  • Page A links to page B -> implies B is worth looking at.
  • Each page is given an authority depending on how many pages link to it and what the authority of those pages is.
  • Pages that link to many sources share less authority than those that link to few pages. I.E. Authority is shared on each external link, but spamming links lowers the amount of authority shared. Each pages authority is finite.

The algorithm can also be explained through the [see page 37, idea] of a random surfer.

PageRank uses both:

ScoreMeaning
PageRankA measure of the pages authority
IRHow well a document matches the query (eg. through Vector Space Model).

Links to this note