Research And Implementation Of Precise Recommendation Based On Collaborative Filtering Algorithm

With the rapid development of Internet technology,massive data emerges.For Internet users,they are facing a serious problem of "information overload",users can not find the data they need from the massive data.In order to solve this problem,network providers put forward various technologies.Among them,personalized recommendation technology can provide users with customized recommendation results based on users' historical behavior data,which is called precise recommendation.Collaborative filtering algorithm,as a widely used recommendation algorithm in precision recommendation,faces the problems of sparse matrix,cold start and low performance.At the same time,the historical behavior data generated by users are presented in the form of tags.When multiple users tag multiple items,tags produce social attributes,which are called social tags.By combining tag and collaborative filtering algorithm,the recommendation effect can be effectively improved and the precision of accurate recommendation can be improved.In this paper,starting from the actual project,the shortcomings of the traditional Projectbased Collaborative filtering algorithm are supplemented and corrected by tags.Firstly,the improved K-means clustering algorithm is used to cluster tags.Then the concept and calculation formula of tag similarity between items are put forward.The final similarity calculation formula is generated by weighting tag similarity and score similarity between items.Then,the final similarity calculation formula is used as parameter to predict user's score on items and select the formula.Among them,N items with the highest score are recommended to users.Movie Len,a public test set in recommendation domain,is used to test the proposed algorithm to verify the effectiveness and universality of the algorithm.Finally,on the basis of the precise recommendation algorithm proposed in this paper,the precise recommendation engine for Xie Nong project is constructed.At the same time,the data of Xie Nong project are used to verify the specificity of the algorithm.The precise recommendation algorithm proposed in this paper is finally implemented in specific projects to solve the problems encountered in the actual project.The specific work of this paper is as follows:1.The improved clustering algorithm is used to cluster the labels of the project and form a label cluster.The K-means clustering algorithm is selected,and the improved content is the measurement standard of the center distance of the algorithm.The improved algorithm proposed in this paper can adapt to the characteristics of "tag".2.The tag cluster formed by clustering is used to calculate the tag similarity between items.Combined with the item score similarity in the Project-based Collaborative Filtering algorithm,the final similarity calculation formula is weighted.The comprehensive similarity is applied in the prediction score calculation to predict the user's interest value for the unpurchased goods,and Top-N recommendation is selected according to the ranking of interest values.For users.3.For the proposed precise recommendation algorithm,Movie-Len is used to train and test,adjust the parameters of the algorithm,and compare with the traditional collaborative filtering algorithm to verify its performance advantages and prove the universality of the proposed algorithm.4.Design and implement the commodity evaluation and accurate recommendation module of Xie Nong Project,and construct an evaluation system of coexistence of label and score.This evaluation system not only facilitates user evaluation,but also effectively collects user's rating and label data.At the same time,based on the algorithm proposed in this paper,the precise recommendation engine of Xie Nong system is constructed,and the parameters of the precise recommendation model are optimized by using the data accumulated by the current Xie Nong system.