Research On User Recommendation Of Research Network Community Based On User Portrait

Informal academic exchanges in the research network community complement formal academic exchanges,so that research scholars can obtain more comprehensive academic communication channels and cooperate effectively with others.However,information asymmetry and information overload exist in the research network community.How to make research scholars obtain effective information and communicate effectively with others requires personalized recommendations for research scholars.To this end,this article uses personas method to extract the characteristics of users in the research network community to analyze the different needs of users,so as to make personalized recommendations.This article summarizes the personas method and recommendation technology,and uses the personas method to analyze the characteristics of users to make collaborative recommendation to users in the research network community.First of all,based on the AHP-Entropy Weight method to evaluate user influence to extract high-impact users.Next,we extract keywords from high-impact users and corresponding reviewer data from three dimensions: the user’s natural attributes,behavior attributes,and hobbies attributes,different characteristics of the users are analyzed to form personalized tags depicting personal portraits of different users.Secondly,we built two types of recommendation models based on the personas in the research network community: precise collaborative recommendation based on the personas and potential collaborative recommendation based on the personas.In precise collaborative recommendation based on the personas,starting from the user’s active needs,discovering the user’s interest based on the professional relevance of the user’s comment content,the emotional interest of the comment content,and the length of the comment,and quantifying the comment content of the reviewer,among them,the professional relevance calculation of the review content uses the TF-IDF algorithm to construct a professional keyword dictionary for the review content to match,and finally the Comb data integration method is used to obtain the final result for precise collaborative user recommendation.In the potential collaborative recommendation based on the personas,this article starts from the static and dynamic characteristics of users,and explores according to the user’s professional fields,published blog posts,and comments in the user’s natural and behavioral attributes and hobbies attributes.Based on the preferences of users’ potential needs,the LDA topic model is used to calculate the user’s behavior similarity,the collaborative filtering algorithm is used to calculate the user’s interest similarity,and then the linearly weighted method is used tointegrate the above data to make recommendations for potential cooperative users.Finally,this article uses Science Net.cn as an example to construct the personas for users,extract user characteristics,analyze users’ academic needs,and make different communication and cooperation recommendations based on the academic users’ different academic interests or cooperation needs.In order to evaluate the validity of the experimental results,this article compares and analyzes the results of precise collaborative recommendation based on the personas and potential collaborative recommendation based on the personas.It is found that the recommendation results meet the user’s needs,which can help users discover their true cooperation needs and potential cooperation demand,alleviate the situation of information overload,to a certain extent,it can also alleviate the user’s cold start problem.