Research On Ranking Journals Based On Improved PageRank Algorithm

In the 1930s,the British philologist B.C.Bradford proposed the concept of "Core journals".While in the 1960s,American literature metrologist E.Garfield also confirmed the existence of core journals from the aspect of quotations.Along with the discussions of core journals,an evaluation system based on influential factors gradually formed up.In the journal evaluations,measuring a journal's influence power are mainly based on what things as follows:Quantity of published articles,Downloads,Total cited times,Influential factors,Cited half-life etc.Currently,the mainly adopted basis is the influential factor,such as the journal criterion report(JCR)published by WOS,Chinese Social Science Citation Database(CSSCI)published by Nanjing University and so on.Those famous journal citation databases are all following the bibliometrics,and adopting the influential factors to evaluate journals.However,the current evaluation indexes of journal influence power are being widely doubted in the academiccommunity.That is mainly due to the presupposition,measurement methods and influential factors.The traditional citation analysis only considers the number of citations,but ignores the differences brought by quotations among different journals mutually.Generally it is acknowledged that the more citations of a journal,the more influential the journal is.But in fact,considering the citation situation in different journals independently would ignore the literature relations in the science field,so as to ignore the influence brought by the journal itself.Especially in terms of journals that are on the boundary of core journals,the experts' qualitative analysis would be affected by their professional fields and preferences.Thus this paper introduced a new algorithm-PageRank algorithm to sort the journals.This method not only concerns the quotations among journals mutually,but also concerns the different weights among different journals'quotations,which is the journal's self-influence power.In the first chapter this paper introduced the PageRank algorithm from its background of birth and its position in the link mining analysis field.The PageRank algorithm model based on the transition probability was derived from the basic PageRank algorithm which exist the heterogeneity of Rank Sink and Rank Leak.The Markov chain was used to prove the existence and uniqueness of the eigenvector of transfer matrix with eigenvalue of 1,and also proved the definiteness of its feature vector.The power method for solving PageRank algorithm was also introduced.The second chapter mainly introduced the randomized reduction algorithm for solving PageRank algorithm,put forward the idea of stochastic reduction,and further expounded and compared the Kaczmarz algorithm with Randomized Kaczmarz algorithm.The third chapter is the empirical analysis.This is about ranking the influence power of 124 kinds of journals about Statistics and Probability in JCR based on PageRank algorithm.It was conducted by using postman software to collect the quotation data of those journals in 2016 and using the PageRank algorithm based on Random Reduction algorithm to get the ranking of these 124 journals' influence power.This paper conclude as follows:there is a significant difference between the ranking of journals' influence power based PageRank algorithm and that based on influential factors.