Literature Value Ranking And Author Influence Evaluation Algorithm Based On Hadoop

With the rapid development of science and technology,the output of research results has gradually increased,and electronic media such as literature retrieval systems has gradually become one of the main storage methods of academic achievements.In the face of massive electronic literature data,how to quickly find the literature and author information we need is a current problem to be solved.More accurate ranking of literatures and evaluation of scholars' academic level can reduce the cost of searching literatures and authors,save researchers' time,and enable researchers to quickly find out the research focus and research hotspots in a certain field.The traditional literature ranking and author influence evaluation methods are computationally complex and difficult to process massive literature data.In order to more reasonably rank the literature and influence the authors in the massive academic data,this paper uses the data processing capabilities of the Hadoop cloud computing platform to research on these two issues based on the MapReduce computing framework.The main work includes the following two aspects:(1)A literature ranking NTMP(Notave-Timefactor-Mapeduce-Pagerank)algorithm is proposed.It solves the problem that the traditional PageRank algorithm ignores the influence of publication time on the process of literature value change.The NTMP algorithm is based on the MapReduce computation framework.It adds the influence of time factor on the literature value ranking and optimizes the distribution way of NTMP value in the iterative process.Experiments show that the NTMP algorithm is more reasonable in evaluation results than the traditional PageRank algorithm.When performing large-scale dataset calculations,it takes less time than the PageRank algorithm.(2)Based on the NTMP algorithm combined with other features related to the author's influence,a multi-feature author influence evaluation algorithm MFAI(Multi Feature Author Influence)is proposed.The algorithm evaluates the author's influence by calculating the value of the author's literature,the influence of the author's research institution,and the author's cooperation network.In order to avoid the the adverse effect of subjective factors in the calculation process,the grey correlation analysis method is used to calculate the weights of related factors.Using Microsoft academic graph as the data set for verification,the experiment shows that in the evaluation process,the MFAI algorithm can more comprehensively evaluate the author's influence compared with the traditional methods such as H-index and author citation.