Improved Collaborative Filtering Algorithm And System Implementation

With the birth and development of the Internet,the recommendation system has been widely used.Information producers will analyze the interests of users,screen and recommend information,and then get the resources users want to obtain.Collaborative filtering algorithm is a classic algorithm of recommendation algorithm.Collaborative filtering algorithm ignores the content of goods and uses the relationship between goods and users to recommend users.However,due to sparse data,the accuracy of user similarity calculation is low and the problem of inaccurate recommendation still exists.(1)Previous collaborative filtering algorithms ignore the mining of commodity attributes.This paper proposes a collaborative filtering algorithm based on commodity attributes and user attributes.According to the user commodity scoring matrix,the user scores the commodity attribute value;through the analysis of the user's attribute,the Pearson formula is improved;according to the improved user similarity,the user commodity attribute value scoring matrix is filled in;after the user similarity is modified,the scoring prediction formula and the commodity scoring prediction formula based on the user's commodity attribute scoring are weighted to obtain The final scoring prediction formula,and finally recommend top-N products for users.(2)Aiming at the problem of low accuracy of traditional calculation of user similarity due to the lack of consideration factors,this paper improves the formula of user similarity,introduces the concept of user commodity attribute interest degree,forms the proportion matrix through the user interest degree of attribute,and obtains the user similarity degree;proposes the concept of related goods,non related goods,through the related goods and non related goods The concept of product contribution degree is put forward.According to the contribution degree,the user similarity is calculated,and the user similarity obtained by three factors is combined with Pearson similarity to get the improved user similarity calculation formula.The experimental data is movielens data set,and the experimental results show that the improved recommendation algorithm is better than the traditional collaborative filtering recommendation algorithm.