PDF Publication Title:
Text from PDF Page: 077
Algorithm 1 – Compute the derivative of PageRank. 1. Computex(α)asthesolutiontotheoriginalstronglypreferential PageRank problem, (I − αP ̄ − αvdT )x(α) = (1 − α)v. 2. Computez(α)asthesolutiontothestronglypreferentialPage- Rank problem with teleportation distribution x(α), (I − αP ̄ − αx(α)dT )x(α) = (1 − α)x(α). 3. Setz ̃= 1 z(α). α(1−α+αdT z(α)) − e T z ̃ 4. Computeη=eTx(α). 5. Returnx′(α)=z ̃+ηx(α). Program 4 – Strongly-preferential PageRank derivatives. Our PageRank codes for Matlab use row sub-stochastic matrices P. 1 function xp = derivpr(P,alpha,x,z) 2 % 3 % 4 % 5 % 6 % 7 % 8 d 9 zt = z./(alpha*(1-alpha+alpha*d’*z)); DERIVPR Compute the derivative of PageRank Given a PageRank vector x(alpha) and a PageRank vector y(alpha) that satisfy (I - alpha P’)x = (1-alpha)v (I - alpha P’)z = (1-alpha)x we produce the derivative of PageRank at alpha. = 1 - full(sum(P,2)); d = round(d); % compute the dangling vector 10 g = -csum(zt)/csum(x); 11 xp=zt+g*x; A full investigation of algorithms for PageRank derivatives ought to in- clude a discussion about the stability of the algorithms. Such a discussion is a glaring omission of this chapter as the vectors x and z in algorithm 1 will not be accurately computed. As an ode to the missing analysis, let us note that algorithm 1 produces a derivative vector that satisfies eT x′(α) = 0 to machine precision. This property should be useful for a backward stability analysis. Such analyses are usually difficult to conduct and we do not expect this case to be an exception. It is now time to address other properties of the derivative. 3.3 analysis Studying algorithm 1 from the previous section to investigate the deriva- tives reveals a few interesting properties. The investigation begins with taking a Taylor step along the derivative. 3.3 ⋅ analysis 55PDF Image | Instagram Cheat Sheet
PDF Search Title:
Instagram Cheat SheetOriginal File Name Searched:
pagerank-sensitivity-thesis-online.pdfDIY PDF Search: Google It | Yahoo | Bing
Cruise Ship Reviews | Luxury Resort | Jet | Yacht | and Travel Tech More Info
Cruising Review Topics and Articles More Info
Software based on Filemaker for the travel industry More Info
The Burgenstock Resort: Reviews on CruisingReview website... More Info
Resort Reviews: World Class resorts... More Info
The Riffelalp Resort: Reviews on CruisingReview website... More Info
CONTACT TEL: 608-238-6001 Email: greg@cruisingreview.com (Standard Web Page)