知识网络的累积结构与路径长度
Cumulative structure and path length in networks of knowledge
摘要:
科学和技术的一个重要知识层面是其发展的累积程度,即后来的发现在多大程度上建立在先前的发现之上。可以使用网络方法研究累积知识结构,其中节点代表发现,链接代表知识流。对这些研究特别感兴趣的是网络路径和路径长度的概念。从网络增长的Price模型出发,我们导出了从给定的初始节点到网络中每个节点的所有唯一路径的路径长度分布的精确解。我们研究了平均入度和累积优势效应的相对重要性,并实现了一个泛化,其中入度取决于节点的数量。累积优势效应被发现从根本上减缓路径长度的增长。由于所有唯一路径的集合可能包含许多冗余,因此我们还考虑到网络中每个节点的最长路径的子集。由于这种情况更复杂,我们只在简单的上下文中近似最长路径长度分布。在给定长度的所有唯一路径的数量无限增长的情况下,给定长度的最长路径的数量收敛到一个有限的极限,该极限指数地依赖于给定的路径长度。因此,基本的网络性质和动力学特征性地塑造了这些网络中的累积结构,因此在研究这些结构时应该考虑在内。
英文摘要:
An important knowledge dimension of science and technology is the extent to which their development is cumulative, that is, the extent to which later findings build on earlier ones. Cumulative knowledge structures can be studied using a network approach, in which nodes represent findings and links represent knowledge flows. Of particular interest to those studies is the notion of network paths and path length. Starting from the Price model of network growth, we derive an exact solution for the path length distribution of all unique paths from a given initial node to each node in the network. We study the relative importance of the average in-degree and cumulative advantage effect and implement a generalization where the in-degree depends on the number of nodes. The cumulative advantage effect is found to fundamentally slow down path length growth. As the collection of all unique paths may contain many redundancies, we additionally consider the subset of the longest paths to each node in the network. As this case is more complicated, we only approximate the longest path length distribution in a simple context. Where the number of all unique paths of a given length grows unbounded, the number of longest paths of a given length converges to a finite limit, which depends exponentially on the given path length. Fundamental network properties and dynamics therefore characteristically shape cumulative structures in those networks, and should therefore be taken into account when studying those structures.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。