图书馆杂志

图书馆杂志 ›› 2018, Vol. 37 ›› Issue (10): 89-100.

• 情报研究 • 上一篇    下一篇

超网络视角下的科研合作网络节点重要性评估

王 雨 郭进利   

  • 出版日期:2018-10-15 发布日期:2018-10-25
  • 作者简介:王? 雨 女,上海理工大学管理学院,研究员。研究 方向:复杂系统,人类行为动力学。作者贡献:评价 指标及模型的构建,实证数据的收集、处理及模拟分 析,论文的书写及排版。E-mail:15275196106@163. com 上海 200093 郭进利 上海理工大学管理学院,教授,博士生导 师。研究方向:复杂性科学和网络科学,人类行为动 力学。作者贡献:论文的检查和校对,用Matlab进行 编程。 上海 200093

The Evaluation of Node Importance within the Scientific Collaboration Network from Hypernetwork Perspective

Wang Yu, Guo Jinli   

  • Online:2018-10-15 Published:2018-10-25

摘要: 本文从超网络的视角,研究了科研合作中的作者重要性问题。首先,通过综合作者的发文量、期刊的影响因子、论文的时间因子、被引频次以及作者的署名顺序等影响因素,设计了反映作者自身科研价值的指标V,以此表征作者的局部重要性。然后,构造了任意长度的平均路径数矩阵,并根据节点间的距离、平均路径数等影响因素,构建出三个重要性贡献比例矩阵,以此反映全网节点间的相互影响程度,进而得到重要性贡献值F,它可表征作者的全局重要性。最后,综合作者自身的价值V和从其他作者获得的贡献值F,提出了最终的重要度指标。对图书情报领域的 81位作者和855篇论文构建超网络,进行实证分析。结果表明,本文指标能有效地识别出领域内的核心作者。将评价结果分别与传统指标和单一指标进行对比,进一步验证了方法的有效性。

关键词: 超网络, ? 作者重要性, ? 重要性贡献矩阵?, 平均路径数?, 复合被引频次

Abstract: From the perspective of hypernetwork, the paper studies the importance of authors in scientific research collaboration. First of all, by considering the impact factors of journals, time factors and cited frequency of papers, authors’ signature order and the paper quantity, we design author’s own scientific value V, which can characterize the local importance of authors. Then, we construct the average path matrix with arbitrary length. According to the distance between nodes, the average number of paths and other factors, three matrices about importance contribution proportion are constructed, which reflect the influence strength among all nodes. As a result, we obtain the total importance contribution value F that can characterize the global importance of authors. Finally, combined with the author’s own value V and the contribution value F obtained from other authors, the final importance index D is proposed. In the field of library and information, we construct the hypernetwork including 81 authors and 855 papers. The empirical results show that the index proposed can identify the core authors effectively. Moreover, compared with traditional and single indicators respectively, the validity of the method is further verified.

Key words: Hyper network, Importance of authors, Importance contribution matrix, Average number of paths, Composite cited frequency.