利用链路预测揭示合著网络演化机制

发布时间：2018-03-12 15:33

本文选题：合著网络　切入点：演化机制　出处：《情报科学》2017年07期 　论文类型：期刊论文

【摘要】：【目的/意义】掌握合著网络的最佳演化机制及其演变能够更好的进行合著关系预测和推荐,进而为研究团队的人员选择和搭配提供建议和参考。【方法/过程】以共同邻居、到达路径、优先连接和随机游走共4类16种相关性指标表示合著网络演化机制,并运用链路预测的理论和方法系统全面的定量比较不同演化机制的优劣和时序变化,揭示合著网络的最佳演化机制及其变化并解析其成因。【结果/结论】在图书情报领域的实验证实:描述合著网络演化机制的最佳指标为AA(Adamic-Adar);不同时间段的相关性指标的预测准确率具有一定差异但总体趋势保持一致,并且最佳指标所属类别并未改变,表明合著网络演化机制具有较强的稳定性;对多种类别的合著网络演化机制成因及其改进方向进行了分析。
[Abstract]:[objective / significance] mastering the optimal evolutionary mechanism and evolution of the coauthor network can better predict and recommend co-authorship, thereby providing advice and reference for the selection and collocation of research teams. [methods / processes] for common neighbors, There are 16 kinds of correlation indexes of arrival path, priority connection and random walk to express the evolution mechanism of coauthor network, and the theory and method of link prediction are used to compare the advantages and disadvantages of different evolution mechanisms and the change of time series in a comprehensive and quantitative way. The experiments in the field of library and information confirm that the best index to describe the evolution mechanism of the coauthor network is AAADAAM-Adaranic-Adaran.Relativity of different periods of time is indicated by experiments in the field of library and information. The accuracy of target prediction is different, but the overall trend is consistent. And the category of the best index does not change, which indicates that the coauthor network evolution mechanism has strong stability, and analyzes the causes of coauthor network evolution mechanism and its improvement direction.
【作者单位】：南京理工大学经济管理学院;
【基金】：国家自然科学基金青年基金(71503125) 教育部人文社会科学研究青年基金(14YJC870025) 江苏省2011协同创新中心平台“社会公共安全科技”项目
【分类号】：G353.1;O157.5

【相似文献】