基于模糊聚类的歌曲智能推荐方法研究

Study on intelligent song recommendation based on fuzzy clustering

采用模糊聚类算法原理对歌曲进行分类,智能地为用户推荐其所喜爱的同类歌曲。主要方法是首先确定聚类算法对象的属性指标,然后将属性指标依次根据数据模型建立数据矩阵并将矩阵进行差异度分析,以计算对象之间的距离。差异度分析后将矩阵转化成模糊相似矩阵并得出动态聚类图,最后根据动态聚类图,把相似的歌曲放在相同的簇中间,较好的实现了歌曲的动态分类,改变了同类网站采用排行榜给用户推荐歌曲的传统模式。

Songs are classified by the principle of fuzzy clustering algorithm, the similar favorite songs are recommended intelligently for users. The main way is, determining clustering algorithm object＇s properties indicators first and then attribute indicators are based on data model and data matrix difference analysis to calculate the distance between objects. After variance analysis, the analogous matrix will be transformed into dynamic fuzzy similar matrix and dynamic clustering map, finally according to the dynamic clustering map, take the similar songs in the middle of the cluster, realize the song dynamic classification better, and change the traditional model that the similar web sites recommend songs for users by a list.