Abstract: To solve data sparseness problem of recommendation algorithm, this paper fuses the tag to the conditional restricted Boltzmann machines model. It utilizes CRBM??s powerful ability to fit arbitrary discrete distribution to predict the missing scores for the user unevaluated products. Specifically, it proposes the real-valued conditional restricted Boltzmann machine model whose visible units are real value firstly. Then, the TF-IDF algorithm of text classification is used to predict the attitude of the user who applies the tags, which is multiplied by the tag-genome to get the user??s scores for products, which are integrated into the user history rating data. In the conditional layer, this paper incorporates user tagged/untagged {0, 1} vector in the original rated/unrated {0, 1} vector. Finally, the comparative experimental analysis about real-world dataset shows that the proposed method enhances the accuracy of recommendation.

Key words: recommendation algorithm, tag, tag-genome, term frequency-inverse document frequency (TF-IDF), real-valued conditional restricted Boltzmann machine (R_CRBM)

摘要： 针对推荐算法中数据的稀疏性难题，把用户标签融合至实值条件受限玻尔兹曼机（real-valued conditional restricted Boltzmann machine，R_CRBM）模型，利用R_CRBM强大的拟合任意离散分布的能力，预测出用户对未交互商品的评分缺失值。具体来说，首先提出显层单元为实值的R_CRBM模型，接着运用文本分类中的TF-IDF算法预测出用户对所应用过的标签的喜爱度，与标签基因数据相乘得到用户对商品的预测 评分，融合至用户历史评分数据中。R_CRBM条件层在原有评分/未评分{0,1}向量中，融入用户标签/未标签{0,1}向量。通过真实数据集进行对比分析，实验结果表明提出的方法在一定程度上提升了推荐的准确性。

关键词: 推荐算法, 用户标签, 标签基因, TF-IDF, 实值条件受限玻尔兹曼机（R_CRBM）