Research On Recommendation Of Interdisciplinary Scientific Research Collaborators Based On Multidimensional Decision Attributes

With the rapid development of information technology and the popularization of digital research infrastructure,academic models and research organizational forms are undergoing drastic changes.The trend of scientific research breaking through the limitations of a single discipline to cross disciplinary research models is becoming increasingly evident.In order to understand and solve scientific research problems,scholars tend to conduct research through cooperation improve academic output.Previous studies have shown that scholars often choose research collaborators based on three types of motivations: production motivation,economic motivation,and social motivation.This means that scholars need to comprehensively consider various decision-making characteristics attributes,such as other scholars’ disciplinary fields,research directions,academic level,authority and interpersonal networks.However,currently,most scholars choose research collaborators through social circles,participation in academic conferences.Choosing collaborators based on historical cooperation records is not conducive to identifying potential scholars.The complexity and cognitive limitations of the discipline result in scholars missing out on opportunities to collaborate with potential collaborators in scientific research.Although scholars can use research social platforms for academic exchange to identify potential collaborators,the issue of information overload cann’t be ignored.Therefore,this article constructs an interdisciplinary research collaborator recommendation model based on multidimensional decision attributes.The model is based on disciplinary attribute recognition,research interest mining,ability attribute mining,and scientific research role recognition.In the section of disciplinary attribute recognition,this paper selects journals as the disciplinary classification standard,and uses the published volume to identify core scholars within the discipline,and uses the author’s unit to perform homonymous disambiguation.In the research interest mining section,on the one hand,this paper uses the time of publication,the order of authorship and other keywords to weight to characterize the dynamic research interest of scholars.On the other hand,it uses the method of LDA fusion TWE to learn scholars’ research direction feature vectors.In the section of scholars’ ability attribute mining,this paper considers the order of scholars’ coauthorship and the distribution of journals,then uses the domain harmonic h index and the publication equilibrium coefficient to measure scholars’ ability.In the part of scholars’ scientific research role identification,this article judges scholars’ scientific research roles based on the order of their signatures and the description of their contributions in the article.The entropy weight method is used to integrate the above attributes from the three dimensions of discipline association,scholar cognition and team structure,to solve the problems of which discipline scholars are selected to cooperate,how to identify potential collaborators in cooperative disciplines,and how to maximize the matching of cooperation patterns.Finally,this article obtains empirical data from platforms such as CSSCI and CNKI to demonstrate the process of scholars’ mining and recommendation,and verifies the feasibility and effectiveness of the model.