lecture17-linkanalysis-handout-6-per

7 introducon to informaon retrieval sec 213

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: cores as top authori*es. Base set Get in-links (and out-links) from a connectivity server Introduc)on to Informa)on Retrieval S ec. 21.3 Introduc)on to Informa)on Retrieval Sec. 21.3 Itera*ve update Scaling   Repeat the following updates, for all x:   To prevent the h() and a() values from gerng too big, can scale down aler each itera*on.   Scaling factor doesn t really ma\er: h( x ) ← ∑ a ( y ) x x y a ( x ) ← ∑ h( y )   we only care about the rela)ve values of the scores. x y x Introduc)on to Informa)on Retrieval Sec. 21.3 How many itera*ons?   Claim: rela*ve values of scores will converge aler a few itera*ons:   in fact, suitably scaled, h() and a() scores se\le into a steady state!   proof of this comes later.   In prac*ce, ~5 itera*ons get you close to stability. Introduc)on to Informa)on Retrieval S ec. 21.3 Japan El...
View Full Document

Ask a homework question - tutors are online