基于CTM-PMF模型的物品推荐

摘要

In order to overcome the shortcomings of the traditional collaborative filtering method, a new recommendation method based on CTM-PMF (Correlated Topic Model and Probabilistic Matrix Factorization) model is proposed. CTM-PMF model combines the PMF model in collaborative filtering and the CTM model in topic modes. The result of this combination is that both advantages of PMF model and advantages of CTM model are combined, solves the problem of new items. In this method, a factor is also introduced which reflects the change of user's interest on the products bought previously. The algorithm is tested on a self-build product dataset. The experimental results show that the new method has better capacity for recommendation compared to PMF, G-PLSA and UBCF method.%为了克服传统协同过滤推荐技术的局限,提出了一种基于CTM-PMF模型的物品推荐方法.在PMF模型的基础上,引入CTM模型,将PMF模型良好的推荐品质和CTM模型优越的物品表示方法相结合,有效地实现了新物品推荐；通过引入用户兴趣因子,解决了用户对已购买物品的兴趣变化问题.在自建的物品数据集上,利用提出的方法、PMF模型、G-PLSA模型和UBCF方法进行了对比实验,实验结果表明该方法具有良好的物品推荐品质.