Block Structure Convergence Algorithm Of PageRank And Its Improvement

21th Century is the information century. People will do everything with Internet such as working, studying, relaxing and living. At the same time, each part of the society puts their business and information on the WEB;information become numeral and data become webside is the flag of society advance. The problem which is followed by is with so much information, how can we find the one that we need? The best way is use a web tool that can search the information which the user is looking for. Search engine technology is coming out by such requirement. Search engine technology is developed while the development of electric technology which make the information become numeral and data become webside. A famous search engine have to provide the information which user need immediately and exactly. If it can do so, a quick, excellent, usable search algorithm is needed. GOOGLE search engine keeps on top by its PageRank and convergence algorithm. But the convergence algorithm is more importment. The convergence algorithm decides the time and space cost in the procedure of evlauting PageRank Vector straightly. A good convergence algorithm can make the system use the lest time and space cost evlauting the final PageRank Vector. So that the total performance of the search engine will be improved. Today, there is a lot of research on this program. Some of them based on shortening the convergence time and some of them based on improving the veracity of search result. In this paper I design a algorithm which is named block PageRank convergence algorithm in the aim of reducing the time and space spending during the system running. It divided the web into several blocks, then evaluate the blocks' PageRank and the PageRank during the blocks. At last evaluate the final PageRank by multiply the two type of PageRanks. This algorithm reduce the system's space and time spending while the system is running by reducing zero element and zero multiplying. I also show the algorithm's advance in the experiment. Meanwhile, there is something different with the anticipated result. I found the result which is evaluated by block algorithm is much different from the one evaluated by common algorithm. Writer analysis the problem, find that after the web be divided, the links between blocks are updated;meanwhile, the webs which have these links will lost part of their PageRank. In order to resolve this problem, writer improve the algorithm. In each blocks, we make several fictitious links to substitute the links between blocks. After these operation, the interrelated web will have the same links in quantity, so that the final PageRank will more approach the common result. In the experiment,this conclusion has been proved obviously that with the advantage in the algorithm's time and space cost, the result's distortion also reduced greatly.