Research And Application Of Latent Factor Model Recommendation Model Based On Change Point

In the era of speedy development of the Internet,it is more and more convenient to gain information.At the same time,a large amount of data is generated,resulting in "information overload".Recommender systems are one of the valid ways to work out this problem.It has the function of information filtering,which can provide more directional information.In recent years,it has been fully applied in many fields,such as e-commerce,personalized ads,video recommendation and so on.One of the hottest areas of research in recommender systems is Latent Factor Models(LFM).It searches for the feature information of latent factors,by analyzing the known features.However,with the worsening of"information overload",recommender systems still face many challenges,such as poor scalability,data sparsity,cold-start,etc.In order to improve the existing problems such as poor scalability and data sparsity,this paper adds user and item feature information into the LFM.At the same time,taking into account the time information,the change point is integrated into the model to improve the accuracy of the algorithm.The main work includes the following two parts:(1)We study latent factor model recommendation model based on change point(CP＿LFM)and build related algorithm.In this paper,the feature information of users and items is added to the LFM to establish an additive model.Because the parameters of recommender systems may change over time,we consider adding temporal information.We incorporate change point into the additive model and propose CP＿LFM.CP＿LFM continuously finds the optimal change point position and replaces the model parameters in time sequence,so that the model can achieve the optimal recommendation effect.According to the proposed model,the relevant algorithm design is constructed.The experimental results show that the accuracy of CP＿LFM algorithm is better than LFM.(2)The application of movie recommender systems based on CP＿LFM.The data used in this chapter are Douban Top250 movies.The system is implemented using Python,Bootstrap,and Django frameworks.The implementation of the system provides users with the needs of movie recommendations.It is also beneficial to the further application and improvement of the CP＿LFM algorithm.