Efficient Pagerank With Same Out-Link Groups

Traditional PageRank algorithm suffers from heavy computation cost due to the huge number of web pages. In this paper, we propose a more efficient algorithm to compute the pagerank value for each web page directly on the same out-link groups. This new algorithm groups the pages with the same out-link behavior (SOLB) as a unit. It is proved that the derived PageRank is the same as that from the original PageRank algorithm which calculates over single webpage; while our proposed algorithm improve the efficiency greatly. For simplicity, we restrict the group within a directory and define metrics to measure the similarity of the pages in same out-link behavior. We design the experiments to group from 0.5 liked to exact SOLB pages; the results show that such group offers similar rank scores as traditional PageRank algorithm does and achieves a remarkable 50% on efficiency.