PDF Publication Title:
Text from PDF Page: 084
64 4 ⋅ random alpha pagerank 4.4.1 Choice of distribution The first order of business for RAPr is to choose the distribution of A. While choosing a distribution seems more difficult than picking a single value α, the right data makes it easy. The information for the empirical distri- bution of A is present in the logs from the surfer behavior studies discussed in section 4.3.2. This point is illustrated in section 4.5 where we take browsing logs and compute a distribution for α. Picking A based on browsing behavior, however, is yet another choice. It seems correct and natural for the random surfer derivation of PageRank. When the PageRank or RAPr values are used in an application, the metrics of the application should drive the choice of α or A. We return to this point in section 4.8.4. We assume that A has a continuous distribution over [l , r] with 0 ≤ l < r ≤ 1. Two distributions with bounded, continuous support are the uniform distribution and the Beta distribution. In fact, the uniform distribution is a special case of the Beta distribution and consequently, our “default” choice of A is a Beta random variable with distribution parameters a and b, and support [l , r]. To denote this, we write A ∼ Beta(a, b, [l , r]). The probability density function for this random variable is ρ(x) = 1 (x − l)b(r − x)a (r−l)a+b+1 Beta(a+1,b+1) . (4.8) It reduces to a uniform distribution when a = b = 0. Later, we will derive our algorithms in the most general settings possible, but all computations are done with some version of the Beta distribution. Section 4.5 presents an empirical distribution strikingly close to a Beta.PDF Image | MODELS AND ALGORITHMS FOR PAGERANK SENSITIVITY
PDF Search Title:
MODELS AND ALGORITHMS FOR PAGERANK SENSITIVITYOriginal File Name Searched:
gleich-pagerank-thesis.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)