基于协同过滤和对象模型的音乐推荐系统研究

Research on Music Recommender System Based on Collaborative Filtering and Object Model

传统的音乐推荐方法在面对诸如新用户,新音乐以及评分稀疏等问题时处理的不够好。基于此,论文提出一种在使用协同过滤算法的基础上融合对象模型的方法。为了解决新用户问题,使用用户画像结合基于用户的协同过滤方法来预测未知评分;为了解决评分稀疏和新音乐问题,未知评分由音乐标签的评分来初始化,然后使用基于商品的协同过滤方法来挖掘用户的偏好。实验结果表明该方法在均方根误差方面比传统的方法具有更好的推荐效果。

Traditional music recommendation methods are not well handled in the face of issues such as new user,new mu⁃sic,and rating sparsity.Based on this,this paper proposes a method of associating the object model with the collaborative filtering algorithm.In order to solve the new user problem,the user profile is combined with the user-based collaborative filtering method to predict the unknown rating.In order to solve the rating sparsity and new music problem,the unknown rating is initialized by the rat⁃ing of the music tag,and then the item-based collaborative filtering method is used to retrieve user preferences.The experimental re⁃sults reveal that,this proposed method performs more promising than the compared methods in terms of Root Mean Squared Error(RMSE).