中文摘要：

针对现有推荐算法在处理海量数据时效率和精确度低下的问题,提出一种将基于内容的推荐算法与基于项目的协同过滤算法相结合的新的混合推荐算法。首先引入符号数据分析方法,使用模态符号数据对项目建模,同时针对评分矩阵的超高维及稀疏性问题加入非负矩阵分解算法,对项目的特征矩阵进行有效＂平滑＂,以此为基础计算项目之间的相似性,进而完成混合推荐。基于MOVEILENS数据的实验结果表明,本文提出的混合推荐算法与传统的基于项目的协同过滤算法相比,在相似性计算上具有更高的效率,同时在应对数据稀疏性及新用户问题时,具有更高的推荐精度。

英文摘要：

The efficiency and accuracy of traditional recommendation algorithms decrease significantly when dealing with massive data. To overcome these disadvantages we propose a hybrid algorithm combining content-based algorithms and user-based collaborative filtering. To better cope with the increasing dimensionality and sparseness of the feature matrix, we construct the user profile by symbolic data and smooth the feature matrix of items with non-negative matrix factorization. The experimental results using the data set of MOVEILENS indicate that the new algorithm is more efficient and accurate compared with traditional user-based collaborative filtering.