Analysis And Modeling Of Double-layer Networks

With the continuous deepening of the globalization trend and the continuous development of technology,collaboration networks,citation networks and field networks have gradually formed large complex systems,and the demand to promote scientific development through the research of academic networks is becoming more and more urgent.The evolution of collaboration networks and citation networks is one of the research hotspots.Previous research focused on the independent evolution of collaboration networks and citation networks.The phenomenon of mutual influence when collaboration networks and citation networks evolve together has not been widely observed.This article is devoted to mining the coevolution phenomena of collaboration networks and citation networks,and discovers the universal law of network evolutions,and explains the coevolution phenomena through the establishment of adaptive coevolution models.In this paper,under the theoretical framework of multi-layer networks,we take the collaboration network and the citation network as layers of the double-layer network and establish a heterogeneous double-layer network.We describe the adaptive coevolution phenomena in the network.The adaptive coevolution phenomena refer to the important interactions between scientists' collaboration and scientists' citations,and the interactions have obvious regularity on the time scale.This paper proposes serval time correlation measures to quantitatively describe the coevolution phenomena of the network.Based on the existing models,this paper proposes an adaptive coevolution model of heterogeneous double-layer networks,and introduces a new inter-layer interaction mechanism of collaboration and citation.The network generated by the model not only fits multiple measures of the real network well,but also completely reproduces the adaptive coevolution phenomenon in the heterogeneous double-layer network.This paper also compares and discusses the important parameters of the model,and verifies the necessity and effectiveness of mechanisms.Finally,this paper introduces several algorithm implementation of the preferential attachment mechanism,and proposes a brand new and efficient implementation of incremental preferential attachment mechanism.In addition to the analysis of the evolution of heterogeneous double-layer networks,this paper also studies academic field evolutions.This paper uses a real scholarly dataset from Microsoft Academic Graph to establish a field network and defines three measures that are closely related to academic field evolutions.These three measures are the cumulative number of papers,the relative proportion of papers and the relative proportion of degrees.By analyzing the trends of the important measures of computer science fields,this paper describes the birth,growth,saturation,and shrinkage phases of fields in detail.For the first time,this paper unifies the collaboration network and the citation network into a heterogeneous double-layer network,and describes the coevolution phenomena of the network,which not only open a new mind of the research for the coevolution of the collaboration network and the citation network,but also help people in-depth understanding of the collaboration pattern among scientists,the reference pattern among scientists and the interaction between collaboration and citation.This paper also establishes a coevolution model of heterogeneous double-layer networks and introduces a new inter-layer interaction mechanism of collaboration and citation,which provides an innovative theoretical network model for explaining the inter-layer interactions of double-layer networks.The detailed description of different stages of field evolutions in this paper reveals the universal law of field evolutions and helps people to discover research frontiers and hot topics.