Research On Personalized Recommendation Algorithm Based On User Interest

In recent years,with the rapid development of Internet big data,information resources have witnessed explosive growth,and the phenomenon of information overload has become more and more prominent,which makes it difficult for users to find the information they need from the massive information resources.The emergence of recommendation systems has solved this problem,as they can help users identify useful information and meet the personalized needs of different users.They are widely used in fields such as e-commerce,tourism,and multimedia.The core of recommendation system is recommendation algorithm.Collaborative filtering is the most widely used recommendation algorithm at present,but there are still some problems such as data sparsity,scalability and low recommendation accuracy.To solve the above problems,this paper fully studies the collaborative filtering recommendation algorithm around user interest preferences,mainly including the following work:(1)A collaborative filtering recommendation algorithm based on fuzzy clustering and user interest is proposed.The user’s interest preferences are usually reflected in the user’s rating value for the item,but a single user rating value cannot fully represent the user’s level of preference.Based on the user’s rating value,this paper uses the user’s category access rate to represent the user’s preference,defines the user’s local preference and global preference,and comprehensively obtains the user’s interest matrix.The fuzzy C-means clustering algorithm based on particle swarm optimization is used to cluster the user’s interest preferences,and the average value of the user’s rating on the item with similar preferences is filled into the user item rating matrix,which alleviates the sparsity of user rating data.Finally,the algorithm comprehensively considers user item ratings and user interest preferences to calculate user similarity,and introduces an item type penalty factor to further improve the accuracy of user preferences.Experiments show that the algorithm proposed in this paper can effectively improve recommendation accuracy,and MAE value is reduced by 8.9%compared with the traditional recommendation algorithm.(2)The concept of user trust relationship has been introduced to improve user trust.This paper calculates the direct trust of users based on their trust relationships.The indirect trust of users is calculated from the similarity of their item ratings,taking into account the influence of user mutual evaluation and rating trust.Finally,the trust weight matrix of users is obtained comprehensively.In addition,in view of the sparsity of the user weight matrix,this paper introduces the user relative activity factor into the user indirect trust,and builds a more dense user weight matrix.Experiments show that the improved user indirect trust improves the accuracy of recommendations and alleviates the sparsity of the user weight matrix.(3)According to the improved user trust,a collaborative filtering recommendation algorithm based on user trust and interest distribution is proposed.In this paper,user interest is introduced into the algorithm and user interest model is built.Firstly,the item is classified,and the distribution of users’ interest in the item is calculated according to the obtained category attributes,and then the user’s common preference factor is obtained.Finally,this paper distinguishes the size of the user’s trust value,defines the user’s trust threshold and the user’s interest threshold,and integrates the user’s trust and user’s interest to obtain the weight value of the user.Experiments show that the improved recommendation algorithm improves the recommendation accuracy and has better performance.