论文部分内容阅读
【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。
[Purpose / Significance] To master the best evolution mechanism of co-author network and its evolution can better predict and recommend co-relationships and provide suggestions and references for the selection and collocation of research teams. [Method / Process] A total of 16 categories of co-related network evolution mechanisms, including common neighbor, route of arrival, priority connection and random walk, are used to represent the evolution mechanism of network. The theory and method of link prediction are used to systematically and comprehensively compare different evolution mechanisms The merits and the changes of timing, revealing the best evolution mechanism of co-author network and its changes and analyzing its causes. [Results / Conclusion] Experiments in the library and information field confirm that the best indicator describing the evolution mechanism of co-author network is AA (Adamic-Adar). The accuracy of the correlation indexes in different time periods is different but the overall trend is consistent , And the category of the best indicator did not change, indicating that the co-author network evolution mechanism has a strong stability; the causes of the evolution mechanism of co-author networks and the direction of improvement are analyzed.