An Effective Accelerating Iterative Algorithm For Computing PageRank

In recent years PageRank problem has become the core problem of the websearch engine,many algorithms were designed to solve the problem.Based on the MSI algorithm,we propose a new algorithm to accelerate solving the PageRank problem in this paper.The algorithm is called MMSI algorithm,which is a generalization of the MSI algorithm by applying the first multi-step matrix splitting iteration and the second matrix splitting to accelerate the algorithm.We derive the algorithm in detail and prove the convergence.Numerical experiments also show the new algorithm is better than the MSI algorithm and the MPIO algorithm.In chapter 1,we briefly introduce the research background and current situation.In chapter 2,inspired by the MPIO algorithm,we propose the MMSI algorithm based on the MSI algorithm,and prove that MMSI algorithm converges faster than MPIO algorithm under some certain conditions.In chapter 3,numerical experiments show that the new MMSI algorithm proposed requires fewer iterative steps and less CPU time than the MSI algorithm and the MPIO algorithm,and the MMSI algorithm is more efficient than the MPIO algorithm.In chapter 4,we give the summary and point out how to further the work.