Research On The Precision Recommendation Method Of Peer Review Experts Based On User Portrait

In recent years,academic misconduct have emerged one after another,triggering the whole society's reflection on issues such as academic integrity and peer review.As an important means of scientific and democratic review process,peer review has played an important role in the selection of excellent results,the optimization of resource allocation,the direction of scientific research and the selection of talents.At present,the selection of evaluation experts mainly depends on the subjective judgment of the editorial department,which not only wastes its working time,but also has many hidden dangers.Based on this,this paper designed a more effective recommendation method for peer review experts by using relevant technologies such as user portrait and recommendation system,so as to provide theoretical and methodological support for improving the efficiency of peer review organizations,improving the satisfaction of both "supply and demand",purifying the academic environment,and optimizing the allocation of academic resources.First,designed the expert user portrait model,from the demographic attributes,academic field label tag,auditing behavior,social relationship experts to portrait the four dimensions,to "tag" of experts,the calculation method of design of each dimension gives different tag corresponding weights,and associated values,and using the method of production and inventory management related to subsequent updates of portraits,experts in dynamic portrait model is set up,secondly,through in-depth study recommendation algorithm was proposed based on collaborative filtering recommendation,a picture of the users of different academic experts and evaluation object tag weight as a reference factor score matrix similarity calculation,at the same time value and build a tag preference based on tag preference score matrix to calculate the similarity of the two kinds of similarity calculation through can adjust the parameters of linear mixture,to improve the collaborative filtering recommendation algorithm.In addition,SVD algorithm is adopted to optimize data sparsity and other problems,so as to design a more effective peer review expert recommendation method based on user portrait.By crawling a large number of experts and paper data on the network,a small review expert database was established,and the recommended methods proposed in this paper were verified.The experimental results show that the improved recommendationalgorithm can effectively carry out expert recommendation and has significant improvement in accuracy,recall rate and comprehensive evaluation index.