Incremental weighted bipartite algorithm for large-scale recommendation systems

Authors: HAIHONG E, JIANFENG WANG, MEINA SONG, QIANG BI, YINGYI LIU

Abstract: Personalized recommendation systems are a solution for information overload. In the real recommendation system, the data are dynamic; thus, the real time of the system will be seriously affected if we recalculate after every change, and, in time, sensitive situation of the delayed updated calculation will affect the accuracy of the recommendation. The incremental calculation methods of collaborative filtering, clustering, singular value decomposition, and bipartite algorithm have achieved some progress; however, there is limited research on the incremental computation of the weighted bipartite algorithm. This paper proposes a personalized recommendation algorithm. In this algorithm, the rank of the recommendations is effectively influenced by the rate and times of the same choice. This paper uses the borrowing records of students in college libraries to model the recommendation system and achieve a rank of personalized recommendations for college students. This system, combined with the autonomic learning resources platform, will improve student learning efficiency. \vs{-1mm}

Keywords: Recommendation system, incremental weighted bipartite algorithm, personalized point of learning interests

Full Text: PDF