面向科研关系网络的发现算法研究

Research on Discovery Method for Scientific Research Relationship Network

采用数据挖掘等技术手段自动发现科研关系网络中蕴含的社区结构,对于有针对性地组建科研队伍、维护项目申报和职称评审等程序的公平公正具有重要意义。现有的科研关系网络发现方法大多存在算法复杂度高、片面依赖论文合著和项目合作关系等问题,因此本文基于经典的PageRank算法,提出一种面向科研关系网络的发现算法。首先刻画科研关系模型,建立科研关系网络,然后以单个科研人员为核心通过对局部网络进行挖掘分析,快速发现其所在的社区,不仅降低了社区发现的复杂度,而且提高了所发现社区的可用性,对我国科技管理工作起到一定的推动作用。

Using data mining and other technical means to automatically detect the community structure contained in the scientific research relationship network is of great significance for the targeted establishment of scientific research teams,the maintenance of fairness and impartiality of project declaration,title evaluation and other procedures.Most of the existing discovery methods for scientific research relationship network have problems such as high complexity,one-sided dependence on paper co-authorship and project collaboration.Therefore,this paper presents a discovery method for scientific research relationship network,which is based on the classical Page Rank algorithm and considers more relationships among scientific researchers.We first describe the relationship model of scientific researchers and then establish the network among them.After that by mining and analyzing the local network with a single scientific researcher as the core,we can quickly find the community in which the scientific researcher is located.Our method can not only reduce the complexity of community discovery,but also improve the availability of the discovered community.We believe that it will play a certain role in promoting the management of science and technology in our country.