A Survey and Comparative Study of Different PageRank Algorithms

Abstract

Searching the World Wide Web is an NP complete problem with sparse hyperlink matrices. Thus searching the significant search results is a challenge. Google's PageRank attempted to solve this problem using computing of principle Eigenvalues termed as PageRank vector. After this, a number of techniques were developed to speed up the convergence patterns of pages in the PageRank algorithm. This is paper, we have reviewed a number of of PageRank computation techniques. The main objective of all these techniques is the convergence rate along with space and time complexities. In this paper, a comparative study is presented among Standard Power method, Adaptive Power Method and Aitken's method using SNAP Google web pages dataset.

Google pagerank algorithm: using efficient damping factor

“…At the start of the loop i=0, by sitting initial PageRank value PR0(v)=1 and PR0(v)=1/n where n is number of total web pages. If there are none in the node's links as in (1). The reason for setting PR0(v)=1 for any other nodes with in links is on assumption that the web pages has at most one in links [16].…”

“…The age that we are living in, which is the age of informati on by many thinkers and scientists. This is for sure growing especially because now, almost everyone has the possibility to generate and share content with others [1]. Modern web search engines can display relevant, popular, and accurate pages towards the top of the query results list.…”

Recursive Join Processing in Big Data Environment

“…Three Mappers are created as shown in Figure 4 There are some types of join queries such as two-way join, multi-way join, or recursive join. Of the three types of join above, recursive join is a fairly complex query with a large cost and is applied in many fields that require repeated calculations such as PageRank [17], Graph mining [27], and Friend graph [20,26] in social networks. Recursive join is also known as "fix-point" join, with repeated operations until the iteration no longer produces new results (Initialization) A(x, y) = K(x, y), (Iteration) A(x, y) = A(x, z) 1 K(z, y).…”