Collaborative Filtering Recommendation Algorithm Research Based On Multi-Feature User Clustering

In recent years,application of recommendation algorithm in real life such as A-mazon,Taobao is getting universal,but it's not perfect yet.A few problems need to be solved such as sparse data and low recommended accuracy.Collaborative filtering is a mature algorithm in the recommended systems,but there are still some problems such as the calculation of similarity and the problem of searching for neighbors of target users.In view of the above issues,this article starts from the following aspects:(1)First,this paper proposes collaborative filtering recommendation algorith-m based on predictions and evolutionary clustering.In order to obtain denser and reliable score data,score matrix is pre-processed with normalization and dimension re-duction.Based on these processed data,clustering principle is generated and dynamic evolutionary clustering is implemented.Besides,the search for the nearest neighbors with highest similar interest is considered.A measurement about the relationship between users is proposed,called user correlation,which combines the satisfaction of users and the potential information.In each user group,user correlation is applied to choose nearest neighbors to predict ratings.The proposed method is evaluated using the Movielens dataset.Diversity experimental results demonstrate that the pro-posed method has outstanding performance in predicted accuracy and recommended precision.(2)Second,the user and project properties are considered to construct the user adjacency matrix.After the definition of similarity of user attribute and difference of user preference item category,the two similarity values are linearly combined.In order to narrow down the search scope of the nearest neighbors of the target users,we use Kmeans algorithm to group the users,and by this doing,we can predict scores within the group.Then,the final score is weighted based on the scoring and attributes.This is to say,the impact of multiple factors is fully considered,such as rating information and attributes.At the same time,the experimental part also verifies the effectiveness and feasibility of the proposed algorithm in prediction and recommendation accuracy.