Research And Application Of E-commerce Personalized Recommendation Algorithm Based On Mobile Platform

With the advent of the era of large data,the data on the platform is becoming more and more overwhelming,which is particularly prominent in the field of e-commerce,facing such a large number of goods but make it easier to spend more time looking for the target,This phenomenon is called "information overload".How to effectively solve this problem caused the exploration from a lot of scholars and personalized recommendation algorithm came into being.This phenomenon in the mobile Internet era has become more serious,the study of e-commerce personalized recommendation algorithm can better help mobile users quickly find the goods they need.In many personalized recommendation algorithms,collaborative filtering recommendation algorithm has the best development prospects and applicability,but there are still problems such as sparse data,cold start and poor scalability.In this paper,the recommended algorithm is thoroughly explored and researched for the main hotspots of E-commerce recommendation system and collaborative filtering recommendation algorithm.Firstly,the recommendation principle of collaborative filtering recommendation algorithm is studied,and various factors influencing the accuracy of recommendation system are analyzed.Based on the above research results,an improved algorithm based on singular value decomposition and BP neural network prediction is proposed.The algorithm uses the singular value to decompose the user-project scoring matrix,which effectively reduces the sparsity of the matrix.At the same time,based on the singular value decomposition,the BP neural network is used to predict the unqualified target project.The advantage is that the use of the average score in place of the effect of the singularity can be avoided,so that the recommended accuracy has been greatly improved.On this basis,the goods will be recommended to the target user,with a similar neighbor list obtained from the similarity calculation.The core idea of this algorithm is to reduce the user-project score matrix dimension,effectively alleviating the sparseness of data,and making the recommendation more accurate.In order to evaluate the effectiveness of the algorithm,this article verifies through the design of a series of experiments.The experimental data is based on the standard Movie Lens data set.The experimental design takes into account the effects of different sparseness and different neighbor users on the performance of the algorithm.It indicates that the improved algorithm in this paper promotes the accuracy of the algorithm in both cases by approximately 4.5%.Finally,in the combination of Android platform for electric business personalized recommendation system design and implementation,the specific needs of the analysis of electricity with mobile platform,according to the preferences of different users to select the appropriate goods to recommend.