Research An Design On Collabrate Network Based Talent People Recommendation System

Nowadays,cooperation and communication between colleges and universities have become one of the major output ways of academic achievements.Academic cross-school research collaboration not only promotes communication in the academic circle,but also the flows and exchanges of talents among universities has become increasingly frequent.A large number of cooperation activities have produced a large number of cooperation results,and it has led to a large amount of academic exchange data.A huge network of collaborators has also formed quietly.How to excavate valuable information from it has become a hot topic of research.Talent recommendation is one of the most important topics.Talent recommendation,that is,colleges and universities provide valuable reference suggestions to select candidates when they introduce sophisticated talents.When introducing talents,colleges and universities not only consider the academic level of scholars,but also consider the cooperative relationship between them and their own researchers.Therefore,it is necessary to analyze the network of collaborators to find out the laws.In the homogenous collaborator network,there are only author nodes,and the joints between nodes are cooperative relations.On this basis,this thesis constructs a heterogeneous partner network,that is,there are multiple types of nodes and connections in the network.The main research issue in this thesis is to find indicators of the strength of association between metrics in the network of collaborators and to recommend potential academic talent for universities.In summary,the main work of this thesis is as follows:(1)First,the SCI paper data of the WOS retrieval platform is crawled,cleaned,denoised,and processed to obtain an experimental data set for constructing a homogeneous/heterogeneous cooperation network.(2)This thesis briefly introduces the evolution of the research methods of the side weight in the cooperation network,and determines the collateral network side weight by introducing cooperation effects.A link-based recommendation model is proposed based on the homogenous cooperation network,and the definition and calculation method of the cooperation trust degree are given.The constraints of different cooperation relationships are analyzed,and the definition of the recommendation probability is given under this constraint.(3)Based on the heterogeneous cooperation network,a PathSim algorithm based on journal partition is proposed.Experiments show that the algorithm's result is more in line with the actual situation.Taking into account the semantic information of the meta-path,this paper measures the similarity between scholars synthesizing two meta-paths.(4)An authoritative ranking algorithm based on dual-type information network is used to calculate the authoritative value of scholars in institutions,and a talent recommendation model based on authoritative ranking of scholars is proposed.And a analytical case is given to compare the recommendation algorithm of the two network types.The results show that the recommendation quality based on heterogeneous cooperation network is better.Finally,the system design is given based on the talent recommendation model provided above.