Many-to-many Collaborator Recommendation Based On Matching Markets Theory

Scientific research cooperation is necessary in the field of innovation and the improvement of output,and helping scholars find suitable collaborators is essential for achieving higher scientific success.Traditional methods fail to focus on many factors such as the willingness of scholars to choose,the limited time and energy of scholars in scientific research cooperation.The recommendation of real academic collaborators should be a process of matching each other at will.Therefore,based on the matching market theory,this paper explores the implementation methods of scholars to select each other for cooperative matching.At present,the focus of some studies in the field of recommendation of academic collaborators at home and abroad aim to provide a list of recommended candidates for the goal,regardless of the scholars' willingness to choose and the times of recommending them.Hence,it is still a greater challenge to explore the cooperative matching method concentrating on the will of scholars and the limitation of scholars' cooperation in scientific research.This paper not only proposes a many-to-many collaborator recommendation method based on matching market theory,but explores the optimal cooperation matching in the process of collaborator recommendation.On the one hand,we mine scholars' correlation from academic big data with network representation and other methods,and applies matching theory to the recommendation process of academic collaborators.The matching market theory will explore the optimal matching in the cooperative matching process.On the other hand,the demand for scholars in terms of the number of collaborators is confirmed with the help of analyzing the number of scholars' collaborators each year.Simultaneously,the correlation among scholars is evaluated for cooperation and matching with some factors such as the co-authored information,research content,and participation information in academic big data.In the matching process,the most popular model and the least-cost stride model are proposed to explore the optimal matching solution method in the matching process.The benefits achieved by cooperative matching are evaluated by means of the definition method in matching market theory.This paper validates the proposed method based on Microsoft dataset and DBLP dataset.Experimental results showcase that the application of matching theory in the recommendation method of academic collaborators effectively optimizes the recommendation results and improve the accuracy of recommendation to some extent.