Research And Implementation Of Recommendation System Based On Hierarchical Enhancement Weighted Similarity Algorithm

In order to pursue the spiritual life with higher quality,most people choose to listen to songs as a way to release their psychological pressure and enjoy relaxing life.Now a variety of music online-sites and mobile-phone software emerge in an endless stream,basically to meet the psychological needs of users.Personalized music recommendations can also make users feel surprised.The mutual integration of search engine and recommendation module promotes the rapid development of personalized music-recommendation system.With the gradual improvement and optimization of the recommendation algorithm,the recommendation results can be dynamically updated with the user's behavior,which is convenient for users to quickly find the songs that meet the requirements in the information-overload environment.Currently,there were many researches on recommendation algorithms,but some recommendation algorithms still had the problem of low accuracy in the link of song recommendation,and could not effectively distinguish recommendation categories according to users' interests.In order to improve the recommendation performance of the recommendation system,improving the accuracy of the recommendation algorithm was the key point to realize this technology.Accuracy was closely related to similarity between objects.At present,cosine similarity and Euclidean distance formula were popular similarity calculation formulas,and relevant experts had carried out a lot of research in different fields by using the above methods.However,the existing research methods rarely considered the influence of user behaviors on user preferences when calculating similarity,and no effective measures were taken for similar behaviors and different behaviors of users,which made the system unable to achieve high precision in recommendation-performance requirements.In order to solve the above problems,weighted similarity algorithm based on hierarchical reinforcement is proposed in this paper.In this algorithm,different behaviors of users are divided into hierarchies,and different weight values are assigned according to the difference between them.For the two users with different behaviors,the similarity value is weakened on the original basis,while for the two users with similar behaviors,the similarity value is enhanced on the original basis to play the role of gaining measurement.The simulation results show that the algorithm has better similarity results with clustering data sets.Then,the weighted similarity algorithm based on hierarchical reinforcement is applied to the system design and development.The nearest neighbor object obtained by this algorithm is combined with the user-based collaborative filtering recommendation algorithm to complete the recommendation process of the whole algorithm,and the music-collection recommendation system is designed.The system uses SSM framework in the back-end development and EasyUI framework in the front-end development.According to the type,language,singer,popularity and other aspects,users are provided with different functional modules.For the personalized music recommendation part,this paper uses the proposed algorithm to calculate the score of songs for users and recommend appropriate songs.Finally,the system of the related functional modules is tested to verify the stability and feasibility of it,and it shows that the system has higher accuracy of the recommended results.