An Algorithm For Maximizing The Global Influence Of Academic Network

Papers are the gathering place of modern human new ideas and knowledge.It is of great significance for academic research to search out the most influential papers from the vast collection of papers.At present,through improving the propagation model and optimizing influence maximization algorithm,researchers have made fruitful achievements in the related research on the problem of social network influence maximization.The research of academic network is based on the existing social network research and combines the features of academic network to calculate the influence between nodes.But when calculating the global influence of nodes,we still need to use greedy algorithm with high time complexity,which greatly reduces the execution efficiency of the algorithm.In order to overcome the shortcomings of previous studies,this paper proposes a new algorithm for calculating the global influence of academic network nodes,which combines the features of acdemic network.The algorithm mainly includes:1)Expanding the linear threshold model.In the process of influence propagation,the influence of all nodes is propagated downward one hop at a time,and new propagation influence are obtained,and all influence are counted.2)The globalinfluence of a single node is calculated.The direct path influence,indirect path influence and multipath influence among nodes are defined,and a cyclic operation rule is defined.According to the publishing time feature of citation network,a threshold-related influence matrix is constructed and iterated to obtain a single node global influence matrix.3)All nodes are sorted according to global influence,and the first n nodes are selected as candidate nodes to select k seed nodes,so as to avoid the aggregation of nodes with larger influence.This paper takes the real academic citation network data set as the experimental sample,and compares the global-based algorithm for maximizing the influence of academic network with the two benchmark algorithms in terms of activation range and running time.The experimental results show that the proposed algorithm has good performance in activation range and running time.