一种面向共享账号的个性化推荐算法

Personalized recommendation algorithm for shared account

为了解决多用户共享账号情况下,账号内部分用户得不到有效推荐的问题,提出PRASA(personalized recommendation algorithm for shared account)算法,首先利用LDA(latent Dirichlet allocation)主题模型构建项目特征向量,接着利用DPC(density peaks based clustering)算法对项目进行聚类分组,为分组后的每组项目分别进行推荐,对于离群点进行单独处理后产生推荐,保证推荐结果可以覆盖更广泛。实验结果表明,提出的PRASA算法可以有效地为共享账号的用户产生合适的推荐。

In order to solve the problem of multi-user sharing account,this paper proposed the PRASA(personalized recommendation algorithm for shared account)algorithm.Firstly,the algorithm constructed the item’s feature vector by using the LDA(latent Dirichlet allocation)theme model.Then it grouped the items and recommended for each group of items and each of the outliers,the recommended results could be covered more widely in this way.The experimental results show that the proposed PRASA algorithm can effectively generate the appropriate recommendations for the users that shared the same account.