Node Importance Measurement And Community Detection Based On Metapath In Heterogeneous Information Networks

Network widely exists in our real life.After computer modeling,The network can be abstracted into information network,Such as weibo friend information network.Information network is divided into homogeneous information network and heterogeneous information network.The heterogeneous networks contain many types of nodes and edges,which can more truly represent the real-life network.There are a large number of data generated every day in society.Researchers can abstract key information on these large number of data to establish information networks and use relevant algorithms to research these information networks,which can be applied to many aspects such as social security management,public opinion monitoring,commodity recommendation and so on.This paper establishes DBLP academic literature heterogeneous information network from DBLP data set,After that we proposes MPIMM and PathLPA to measure the importance of nodes and community detection in information network.Node importance measurement is one of the dimensions of network research.By effectively measuring the importance of nodes,we can quickly and accurately find the key nodes in the network,This paper adds corresponding measurement rules and introduces the concept of contribution degree of author in his paper to measure the importance of nodes based on authoritative ranking.After the organic combination measurement rules,this paper puts forward the model of node importance measurement-Multipath Iterative Metrics Model(MPIMM),Our Model is compared with several algorithms through experiments.Finally,it is concluded that the measurement model in this paper has good measurement performance.Community detection is also one of the dimensions of network research.It can be applied to many aspects to find better results through good community detection methods.This paper introduces the similarity based on meta-path computation to optimiz the classical label propagation algorithm,This paper puts forward the model of the community detection algorithm-Label propagation algorithm based on meta path(PathLPA).The algorithm is applied to the DBLP academic literature heterogeneous information network to detect the author and achieve good results.