Automatic Recommendation System For Matching Scientific Research Projects With Experts

The automatic recommendation system for matching scientific research projects with experts can match and recommend relevant scientific and technological experts for these institutions according to the scientific research project demand information of scientific research project authorities,local science and technology service institutions.For these organizations,when they encounter technical bottlenecks in the process of development,they can issue relevant demand information through the system.The system can match according to the demand information,recommend related technical experts that meet the needs according to the technical difficulties of the institution,saving the cost of expert search and facilitating the accurate docking of expert wisdom.First of all,this paper analyzes and summarizes the related literature.Based on the existing research,a recommendation algorithm for matching scientific research projects and experts is proposed,that is,the connection between scientific research projects and expert information.When matching,it is necessary to consider not only the consistency between the research project and the research direction of experts,but also the theoretical research ability and practical operation level of experts.According to the practical needs of scientific research projects,the similarity between experts and scientific research projects and the expert ability are calculated to obtain the recommended resultsSecondly,this paper analyzes and processes the text data information according to the data characteristics.According to the system data requirements,designing a web crawler to collect expert information,mostly in the Web of Sciences database,Engineering Village database and science citation database.At the same time,project information,patent information and book information are collected through other data collectors.The collected information is transformed into a structured form.Storing the structured data in the database for later processing and calculation.Thirdly,the LD A theme model and keyword extraction algorithm are used to extract key information from the project information and expert information,and vector representation of the extracted key information is carried out to construct the feature model of scientific research project and expert information feature model.In order to increase the accuracy of keyword extraction,improving the TextRank algorithm by modifying the initial weight of words to ensure that the weight of important keywords is relatively large,which is conducive to better extract keyword information and improve the accuracy of the model.Fourth,in order to improve the accuracy of matching recommendation,a comprehensive analysis is conducted on the theoretical research ability and practical operation level of experts while considering the similarity between the information of scientific research projects and the research direction of experts.In the process of producing the results,this paper analyzes and represents the theoretical research ability and practical operation level of experts according to the papers,works,projects,patents and other information in the expert information,and calculates the correlation index of expert ability.The papers and works published by experts,the projects they participate in and the applications for patents can reflect the research direction and expertise to a certain extent,and the weight can be adjusted in the light of the actual situation of scientific research projects.Finally,this article takes the data of experts in school-enterprise cooperation as the application scenario,uses the actual scientific research projects information and the school's expert information as the data basis to build an automatic matching system for scientific research projects and experts,and design relevant experiments to compare and analyze the experimental results.The final recommendation result proves that the similarity calculation combined with expert ability recommendation algorithm effect is very good,application prospect,and can be further developed and optimized.