Research On Collaborative Filtering Recommendation Algorithm Based On User Interest Shift

Recommendation system provides personalized recommendation on products and services to users.In the traditional recommendation system,the user interest is regarded as constant over time.In fact,the user interest changes over time,which is not been considered in the traditional recommendation system.Hence,tracking the user interest drift becomes a core in designing dynamic recommendation system.However,it is a challenge to find an accurate and effective method that can predict the user interest drift.In order to solve the prediction problem of the user interest drift,this paper adopts clustering and time impact factor matrix to monitor the degree of user interest drift in the class and more accurately predict item's rating.We add a time impact factor to the original baseline estimates and use the linear regression to predict the user interest drift.This paper focuses on user interest judgment and tracking prediction,and puts forward relevant improvements and innovations as follows:1.Studying and analyzing the current situation and deficiencies of the collaborative filtering recommendation algorithm,and constructing the user interest model through the way of the item aggregation and the aging model,which effectively solves the problem of tracking the user's interest drift situation.2.In view of the current situation and insufficiency of the existing clustering algorithm,in order to improve the effect of the clustering algorithm,the existing clustering algorithm is optimized.The clustering result is more suitable for the user interest drift model,which improves the prediction accuracy of the model for user interest drift.3.Considering the user's interest will produce a more complex drift overtime,this paper uses a multilevel aging model to track the user's interest offset.Our comparative experiments are conducted on three big data sets:MovieLens100K,MoviceLenslM and MovieLenslOM.The experimental results show that our proposed approach can efficiently improve the prediction accuracy.