融入用户反馈的网络知识社区好友推荐模型构建研究OA北大核心CHSSCDCSSCICSTPCD

[Purpose/Significance]From the dual perspectives of user similarity calculation and user feedback,high-quality user recommendation is conducive to improving the personalized recommendation level of the network knowledge community.[Method/Process]Firstly,the paper collected and analyzed users'attributes data based on crawler tools and python packages to calculate the comprehensive similarity between user pairs in terms of background,social relationships,and blog information;then calculated the users'core scores and formed a recommendation list based on the similarity matrix and core user ranking;finally,introducing a user feedback mechanism and determine the best top-k friends based on posi-tive and negative feedback results of the user.[Result/Conclusion]The friend recommendation model incorporating user feedback can improve the quality of recommended users,effectively improve the level of friend recommendation,and also verify that friend recommendation is a gradual repair process.