Research On Weighted And Partial PageRank Algorithm

PageRank algorithm is the key to the success of Geoogle search engine,it's also one of the famous algorithms in the web ranking.It is often called web sort,or page ranking.Since the algorithm was put forward,for the reason that it is easy to understand,and the algorithm is simpler to iterate,after nearly 20 years of development,this algorithm has been successfully applied to many fields.But there are also some shortcomings in the original algorithm,so many scholars have made a lot of improvements of the algorithm.The existing weighted algorithms mainly consider the factors that affect the PR value of the nodes,and adjust the weight of the link.But in the existing improvement algorithms,the weight of each link is fixed.Based on the idea of Gauss-Seidel,we think that the new PR value after each iteration should be used in the Construction of probability matrix.Based on this,we propose a weighted PageRank algorithm based on PR value(PRWPR algorithm),which dynamically allocate each link weight.Numerical experiments show that the PRWPR algorithm can allocate more PR value to a few important nodes,and the relevance of web pages is also higher than original PageRank algorithm and WPR algorithm.In the iterative process of the algorithm,the number of web pages increases the difficulty of the algorithm.In 2004,Yen-Yu Chen et al proposed local PageRank algo-rithm,they replaced the global graph with a subgraph,which can reduce the size of the calculation.In 2014,Yuta Sakakura et al improved their work.Numerical experiments show that when using weighted PageRank algorithm on the subgraph constructed by Yuta Sakakura et al,the PR value of the target node is far from the PR value in the global graph.Based on this phenomenon,a weighted influence value calculation method is proposed in this paper.Under the weighted PageRank algorithm,the PR value of the target node obtained by the local subgraph constructed in this paper is more closer to the PR value of the target node in the global graph.It can replace the global graph better.