Research On Influential Node Identification Method In Open Source Software Community Based On Complex Network

The open source software community connects developers all over the world through email.Studying the open source software community network could help us understand the development direction of open source software.High impact developers can speed up the development progress of software projects and high impact software projects can attract more excellent developers.By focusing on high impact developers and software projects,we can effectively predict the future technology trends.However,the structure of open source software community network is complex,including a variety of types of nodes.There are a variety of interactions among single nodes,and there are complex indirect relationships among group nodes.The methods used to identify influential nodes in traditional network could not model the open source software community network.It will cause a lot of network information loss and lead to inaccurate identification of influential nodes by flattening the complex network or transforming it into traditional network.In order to solve the above problems,this thesis models the direct relationship among various nodes in the open source software community network,and proposes a multi-layer bipartite network model,which uses the transitivity of influence among nodes to identify influential developers and projects.For the relationship among group nodes,the heterogeneous hypergraph is used to model the open source software community network,which can reduce the information loss caused by the flattening mapping of node relationship and identify the influential nodes more accurately.The main works of this thesis are as follows:(1)In this thesis,a multiplex bipartite ranking method,called M-Bi Rank(Multiplex Bipartite Ranking Method)was proposed.Firstly,the interaction between developers and software projects is abstracted as a single-layer network.Secondly,the open source software community is modeled as a multi-layer bipartite network according to the coupling of nodes in different interaction relationships.Finally,the influence of various nodes is fused through interlayer transfer relationship,and the influence value that tends to be stable is taken as the final influence of nodes.(2)In this thesis,a method to identify influential nodes based on heterogeneous hypergraph ranking called HHG2Rank(Heterogeneous Hyper Graph Ranking Method)was proposed.This method is mainly divided into two stages: Firstly,build multi-dimensional heterogeneous hypergraph using the follow relationship between developers and the dependency relationship between software projects.Secondly,at the influence transmission stage,the influence of developers and software projects enhance each other through the heterogeneous hypergraph model,and the stable iterative result is the node influence in the open source software community network power.(3)At last,a large number of experiments are carried out to verify the proposed methods in the real open source software community network Git Hub.Experiments show that the proposed methods could identify the influential developers and software projects accurately on the open source software community network.