基于相关性时间会议因素的STVRank算法  

STVRank Algorithm Based on Similarity Time Venue Factors

作　　者：王严鑫 张芳[2] WANG Yanxin;ZHANG Fang(School of Computer Science and Communication Technology,Jiangsu University,Zhenjiang 212013;School of Computer and Infomation Engineering,Nanyang Institute of Technology,Nanyang 473004)

机构地区：[1]江苏大学计算机科学与通信工程学院,镇江212013 [2]南阳理工学院计算机与信息工程学院,南阳473004

出　　处：《计算机与数字工程》2020年第10期2338-2342,2358,共6页Computer & Digital Engineering

摘　　要：如何客观、公正、有效地评价科学文献是文献计量学中长期存在的挑战。传统的PageRank算法在评估科学文献的时候将所有引用视为同等重要,没有考虑到引用关系的多种因素,比如说不同主题间的相关性,时间间隔以及会议影响力等。论文考虑这三种因素对引文网络链接的影响,为它们分配了一定的权重,并结合PageRank算法模型,提出了STVRank算法。该算法通过词向量技术对存在引用关系的科学文献的相关性进行定量分析。真实数据集ANN上的实验验证表明STVRank算法与PageRank,WC,SPRank等排序算法相比,可以提高整个引文网络排名的有效性和稳定性。How to evaluate scientific literature with objectiveness,fairness and effectiveness has been a long-standing challenge in bibliometrics.The traditional PageRank algorithm gives each reference the equal rating of its importance when assessing scientific literature,without taking multiple factors regarding to citation relations into account such as the relevance among different themes,the time interval and the influence of conferences.Based on the impact of the above three factors on the citation network links,a STVRank algorithm assigning weights to each factor combined with the PageRank algorithm model is proposed.This algorithm quantitatively analyzes the relevance of scientific literature containing reference relations based on an advanced technology called word vectors.Experimental verification on ANN,an authentic data set,showes that the STVRank algorithm could improve the effectiveness and stability of the ranking of the entire citation network compared with other ranking algorithms such as PageRank,WC,SPRank.

关 键 词：引文分析 文献计量 PAGERANK算法 词向量 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]