基于合著与引文混合网络的协同评价方法

摘要

分析节点在网络中的位置和关系是网络分析的重要内容,也是为科学评价问题提供了有益的借鉴.从评价的角度,作者和文献之间存在正向的相互影响效应,因此提构建了由作者和文献构成的异质二分网络评价模型,应用PageRank和HITS算法的思想,建立作者和文献的协同评价.基于混合网络模型的协同评价,综合了合作网络和引文网络的结构特征,能够提供更为均衡的度量指标.以情报和图书馆学领域为样本,对模型的参数特征及收敛性进行了分析,通过对比分析说明了算法的有效性.%It is critical content of network analysis to address the position and relationship of nodes embedded in the network, which is also helpful to the issues of scientific evaluation. From the view of evaluation that the forward mutual influence exists between authors and their publications, a heterogeneous network composed of authors and publications was constructed and a co-evaluation approach was proposed by applying the algorithms of PageRank and HITS. The co-evaluation model, based on heterogeneous network, integrates the structure characteristics of co-authorship and citation network, and offers a more balanced metrics, Focused on the sample of library and information science, the empirical research was employed to analyze the rule of the algorithm' s parameter and converge, and approach' s effectiveness was illustrated by the comparison with other alternatives.