基于子图的科学家合作网络家族辨识

Using Network Subgraph to Characterize Scientists’ Collaboration Network Families

摘要:

对科学家合作网络宏观全局结构的研究多表明其小世界和无标度共性特征，而就其微观子图分布特性即网络家族的认知较少。研究基于复杂网络家族辨识的子图比剖面方法、子图组合强度和子图浓度排序方法，系统辨识了若干规模和不同领域的科学家合作网络的家族分类及特征。研究表明，具有宏观全局结构共性的上述科学家合作网络，在微观子图的分布、组合或排序上依次具有一种形式、三种规则及五种模式的特性，因而划分为相应家族类别，并表征了理论型、实验型及理论实验型合作网络家族特征。相关方法可用于系统分析和比较科研合作家族行为特性和演化机理。

Abstract:

Recent research has shown that the small-world and scale-free topological properties of scientists’collaboration network at the macro network level. However, the micro network structure property of these collaboration networks is still unknown. Focusing on micro network subgraph distribution, we generalize the notion of network super family, and apply three methods, i.e., subgraph ratio profile, subgraph combination intensity, and subgraph concentration rank for systematically detecting families in several large-scale scientific collaboration networks. Results show that the subgraph ration, subgraph combination and subgraph sequence of scientists’ collaboration networks are characterized by one form, three rules and five kinds, respectively, thus can be divided into one, three and five kinds of family categories, correspondingly. These family categories can help to define the theoretical, experimental and mixed scientists’ collaboration network features. The research provides subgraph based family comparison methods for analyzing collaboration behavior and evolutionary mechanism of scientific collaboration systems.