On computing PageRank via lumping the Google matrix (original) (raw)

The research focuses on optimizing the computation of the PageRank vector through a technique known as lumping applied to the Google matrix. Traditional methods have faced challenges due to the sheer size of the Google matrix, necessitating alternative approaches to achieve stable results efficiently. This work evaluates the effectiveness of different algorithms for computing PageRank, showcasing that specific lumping strategies not only enhance convergence rates but also reduce computational operations needed, thus offering practical advantages for large-scale web search engines.