Algorithm Research Of Sparse Linear Methods Based On Temporal Behavior And Expert Opinions

Nowadays,the explosive growth of data has caused a serious problem of information overload.How to quickly obtain the effective information required by users from massive data has become a hot research issue in the field of data mining today.The personalized recommendation technology taps the user's hobbies and interests through the user's historical behavior data and provides personalized recommendation services for the target user,which is one of the important technologies to solve the problem of information overload.The main functions of the personalized recommendation system include score prediction and Top-N recommendation.The Top-N recommendation is considered to be more in line with the application needs of user interest recommendation by predicting the user's preference for items.The Top-N recommendation algorithm based on collaborative filtering is widely used.Sparse Linear Methods(SLIM)is the promotion of item-based recommendation algorithm in collaborative filterin g recommendation algorithm.It has the advantages of fast and efficient recommendation.However,there are still problems such as data sparseness,lack of diversity and novelty of recommendations,and user interest drift.To solve these problems,this pape r proposes a Sparse linear Method based on timing behavior and expert opinions using Top-N recommendation.The main work of this article includes :(1)For the problem of novelty of recommendation in the recommendation system,this paper proposes an algorithm that Sparse Linear Method with two-factor penalty.The algorithm introduces the user activity penalty and the project popularity penalty in the SLIM,corrects the project similarity through the user activity penalty factor,and mitigates the long tail effect existing in the recommendation system through the project popularity penalty factor,It improves the recommendation quality and recommend novelty.(2)For the problem of user interest drift in the recommendation system,this paper proposes an algorithm that Sparse Linear Method based on temporal behavior.The algorithm considers time weights in the Sparse Linear Method with two-factor penalty to capture changes in user interest during the time progression process.It improve recommendation accuracy.(3)For the data sparsity problem in the recommendation system,this paper proposes an algorithm that Sparse Linear Method based on expert opinions.This algorithm takes into account the influence of expert opinions on user recommendation in the SLIM model,and designs a new type of expert opinions calculation method.By referring to expert opinions,it improves the accuracy of the recommendation.Experiments on multiple datasets of different sizes and different sparsity show that the proposed algorithm has higher recommendation accuracy and no velty than the existing methods,which effectively alleviates the problems of recommendation novelty,user interest drift,and data sparsity in personalized recommendation algorithms.