Collaborative Filtering Recommendation Algorithm Based On Logistics Popularity And User Trust

With the rapid development of new business such as cloud computing and Internet of things,data information in various fields has been explosively growing,which leads to the emergence of "information overload","information tread" and so on.In order to help users find their favorite products quickly,the recommendation system arises at the historic moment.As the core of recommendation system,recommendation algorithm has become a hot topic of scholars.Among them,collaborative filtering algorithm has become one of the most popular methods,and has been widely applied in recommender systems.However,many recommendation systems only consider the influence of a single factor,which results in low recommendation accuracy and poor recommendation effect.Therefore,it is very important to consider a number of influential factors on the basis of traditional algorithms.A collaborative filtering recommendation algorithm(PPUTCF)based on project popularity and user trust is proposed by introducing two factors that affect the popularity of the project and the degree of user trust on the basis of the traditional algorithm.The PPUTCF algorithm first calculates the popularity of the project and the degree of user trust,then combines the two with the linear weighting method through the fusion factor,obtains the best value of the fusion factor through multiple iterations,and finally obtains the value of multiple evaluation indexes by predicting the change of the nearest neighbor K value of the target,and then according to the evaluation index.The value verifies the performance of the new algorithm.PPUTCF algorithm selects Epinion data sets for experimental verification.The experiment is divided into two steps:first,the project popularity and user trust are introduced,the algorithm model is obtained through the training set and the stable results are obtained.Then the test set is used to get the value of the MAE and other evaluation indexes of the experiment,and the experimental results are compared with the traditional experiments.The results show that the algorithm is better than the traditional algorithm,but the recommended recommendation is less accurate.The two is to introduce multiple factors at the same time:Project popularity and user trust.The algorithm model is obtained through the training set,and the optimum value of the fusion parameter is 0.35.Then the value of the MAE and other evaluation indexes of the experiment is obtained by the test set.Then the experimental results are compared with the traditional algorithm and the previous experiment respectively.The results show that the PPUTCF algorithm is better than the traditional algorithm and the first step of the experimental algorithm,not only has better recommendation effect,but also has higher recommendation accuracy.