现今的推荐算法大多忽略用户偏好和项目属性中的多个特征，而是在单一推荐准则的基础上训练模型进行推荐. 基于多准则的推荐算法通过考虑用户偏好的多个方面，可以为用户行为提供更加准确的预测. 酒店是旅游行业中重要的环节，为了提高旅客体验，实现酒店评分预测，提出了基于矩阵分解与随机森林的多准则推荐算法. 该算法分两步实现，通过矩阵分解训练得出用户对物品在各个准则上的评分特征，然后随机森林学习评分特征预测最终评分. 实验结果显示，相较传统算法，基于矩阵分解与随机森林的多准则推荐算法的准确性和实用价值更高.
Nowadays, most of the recommendation algorithms recommend items similar to the ones the user preferred in the past with a single criterion, ignoring the multiple features of user preferences. By comparison, the multiple criteria based recommendation algorithm can achieve higher prediction rate by considering user preferences in multiple aspects of items. Hotel business is an important part of the tourism industry. In order to enhance the user experience and achieve a better predictive rating system, a multiple criteria recommendation algorithm based on matrix decomposition and random forest is proposed. The algorithm is implemented in two steps, through the matrix factorization models, each criterion gets an eigenvalue to every user and item. Eigenvalues of all criteria combine into an eigenvector. Then, random forest will be trained by the data set consisting of the eigenvector. The experimental results show that compared with the traditional collaborative filtering algorithms, the multiple criteria recommendation algorithm based on matrix factorization and random forest is better in terms of prediction and accuracy.