Extrapolation to speed-up query-dependent link analysis ranking algorithms

Norozi, Muhammad Ali

doi:10.1145/1943628.1943630

Cited by 2 publications

(1 citation statement)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The two pioneer LAR algorithms PageRank (Page, Brin, Motwani, & Winograd, 1998) and HITS (Kleinberg, 1999) are query-independent and query-dependent respectively. They were followed by substantial amount of research (Haveliwala et al, 1999;Haveliwala, Kamvar, Klein, Manning, & Golub, 2003;Haveliwala, 2003;Bryan & Leise, 2006;Rafiei & Mendelzon, 2000;Brinkmeier, 2006;Jeh & Widom, 2003;Drineas, Frieze, Kannan, Vempala, & Vinay, 1999;Kamvar et al, 2003b;Norozi, 2010), to name just a few.…”

Section: Theoretical Background Link Analysis Ranking (Lar)mentioning

confidence: 99%

Faster Ranking Using Extrapolation Techniques

Norozi

2011

International Journal of Computer Vision and Image Processing

Self Cite

View full text Add to dashboard Cite

Extrapolations are techniques in linear algebra that require little additional infrastructure that must be incorporated in the existing query-dependent Link Analysis Ranking (LAR) algorithms. Extrapolations in LAR settings relies on the prior knowledge of the (iterative) process that created the existing data points (iterates) to compute the new (improved) data point, which periodically leads to the desired solution faster than the original method. In this study, the author presents novel approaches using extrapolation techniques to speed-up the convergence of query-dependent iterative methods, link analysis based ranking methods, where hyperlink structures are used to determine relative importance of a document in the network of inter-connections. The author uses the framework defined in HITS and SALSA and proposes the use of different Extrapolation techniques for faster ranking. The paper improves algorithms like HITS and SALSA using Extrapolation techniques. With the proposed approaches it is possible to accelerate the iterative ranking algorithms in terms of reducing the number of iterations and increasing the rate of convergence.

show abstract